CaravanInsurance-DataMining.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predicting Caravan Insurance Purchases"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Using the data compiled by Caravan Insurance, we will try to predict whether someone will purchase Caravan's mobile home insurance. We will use various classification models in combination with techniques to handle unbalanced datasets for our analysis.\n",
    "\n",
    "###### Our analysis follows the following structure -\n",
    "###### 1. Data Exploration and Visualizations\n",
    "###### 2. Association Rule Mining For Understanding Cross Selling\n",
    "###### 3. Classification using Logistic Regression - Unbalanced Data, Undersampled Data, Oversampled Data - Including Business Cost For Each Model \n",
    "###### 4. Classification using KNN - Unbalanced Data, Undersampled Data, Oversampled Data, SMOTE - Including Business Cost For Each Model \n",
    "###### 5. Classification using Naive Bayes - Unbalanced Data, Undersampled Data, Oversampled Data, SMOTE - Including Business Cost For Each Model \n",
    "###### 6. Classification using Bagging - Unbalanced Data, Undersampled Data, Oversampled Data, SMOTE - Including Business Cost For Each Model \n",
    "###### 7. Classification using Boosting - Unbalanced Data, Undersampled Data, Oversampled Data, SMOTE - Including Business Cost For Each Model \n",
    "###### 8. Classification using Random Forest - Unbalanced Data, Undersampled Data, Oversampled Data, SMOTE - Including Business Cost For Each Model\n",
    "###### 9. Summary of Performance Measures For All Models\n",
    "###### 10. Summary of Business Costs For All Models\n",
    "###### 11. Final Recommendation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. Data Exploration And Visualizations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#Read the Data\n",
    "caravan_df_maindataset <- read.csv('C:\\\\Users\\\\Jayant\\\\Desktop\\\\DMPA\\\\Project\\\\Data_sets\\\\Data_sets\\\\Master_dataset.csv')\n",
    "caravan_df_train <- read.csv('C:\\\\Users\\\\Jayant\\\\Desktop\\\\DMPA\\\\Project\\\\Data_sets\\\\Data_sets\\\\Train.csv')\n",
    "caravan_df_test <- read.csv('C:\\\\Users\\\\Jayant\\\\Desktop\\\\DMPA\\\\Project\\\\Data_sets\\\\Data_sets\\\\Test_with_Targer_Variable.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "'data.frame':\t9822 obs. of  86 variables:\n",
      " $ Customer_Subtype                             : int  33 37 37 9 40 23 39 33 33 11 ...\n",
      " $ Number_of_houses                             : int  1 1 1 1 1 1 2 1 1 2 ...\n",
      " $ Avg_size_household                           : int  3 2 2 3 4 2 3 2 2 3 ...\n",
      " $ Avg_age                                      : int  2 2 2 3 2 1 2 3 4 3 ...\n",
      " $ Customer_main_type                           : int  8 8 8 3 10 5 9 8 8 3 ...\n",
      " $ Roman_catholic                               : int  0 1 0 2 1 0 2 0 0 3 ...\n",
      " $ Protestant                                   : int  5 4 4 3 4 5 2 7 1 5 ...\n",
      " $ Other_religion                               : int  1 1 2 2 1 0 0 0 3 0 ...\n",
      " $ No_religion                                  : int  3 4 4 4 4 5 5 2 6 2 ...\n",
      " $ Married                                      : int  7 6 3 5 7 0 7 7 6 7 ...\n",
      " $ Living_together                              : int  0 2 2 2 1 6 2 2 0 0 ...\n",
      " $ Other_relation                               : int  2 2 4 2 2 3 0 0 3 2 ...\n",
      " $ Singles                                      : int  1 0 4 2 2 3 0 0 3 2 ...\n",
      " $ Household_without_children                   : int  2 4 4 3 4 5 3 5 3 2 ...\n",
      " $ Household_with_children                      : int  6 5 2 4 4 2 6 4 3 6 ...\n",
      " $ High_level_education                         : int  1 0 0 3 5 0 0 0 0 0 ...\n",
      " $ Medium_level_education                       : int  2 5 5 4 4 5 4 3 1 4 ...\n",
      " $ Lower_level_education                        : int  7 4 4 2 0 4 5 6 8 5 ...\n",
      " $ High_status                                  : int  1 0 0 4 0 2 0 2 1 2 ...\n",
      " $ Entrepreneur                                 : int  0 0 0 0 5 0 0 0 1 0 ...\n",
      " $ Farmer                                       : int  1 0 0 0 4 0 0 0 0 0 ...\n",
      " $ Middle_management                            : int  2 5 7 3 0 4 4 2 1 3 ...\n",
      " $ Skilled_labourers                            : int  5 0 0 1 0 2 1 5 8 3 ...\n",
      " $ Unskilled_labourers                          : int  2 4 2 2 0 2 5 2 1 3 ...\n",
      " $ Social_class_A                               : int  1 0 0 3 9 2 0 2 1 1 ...\n",
      " $ Social_class_B1                              : int  1 2 5 2 0 2 1 1 1 2 ...\n",
      " $ Social_class_B2                              : int  2 3 0 1 0 2 4 2 0 1 ...\n",
      " $ Social_class_C                               : int  6 5 4 4 0 4 5 5 8 4 ...\n",
      " $ Social_class_D                               : int  1 0 0 0 0 2 0 2 1 2 ...\n",
      " $ Rented_house                                 : int  1 2 7 5 4 9 6 0 9 0 ...\n",
      " $ Home_owners                                  : int  8 7 2 4 5 0 3 9 0 9 ...\n",
      " $ X1_car                                       : int  8 7 7 9 6 5 8 4 5 6 ...\n",
      " $ X2_cars                                      : int  0 1 0 0 2 3 0 4 2 1 ...\n",
      " $ No_car                                       : int  1 2 2 0 1 3 1 2 3 2 ...\n",
      " $ National_Health_Service                      : int  8 6 9 7 5 9 9 6 7 6 ...\n",
      " $ Private_health_insurance                     : int  1 3 0 2 4 0 0 3 2 3 ...\n",
      " $ Income_._30                                  : int  0 2 4 1 0 5 4 2 7 2 ...\n",
      " $ Income_30.45.000                             : int  4 0 5 5 0 2 3 5 2 3 ...\n",
      " $ Income_45.75.000                             : int  5 5 0 3 9 3 3 3 1 3 ...\n",
      " $ Income_75.122.000                            : int  0 2 0 0 0 0 0 0 0 1 ...\n",
      " $ Income_.123.000                              : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Average_income                               : int  4 5 3 4 6 3 3 3 2 4 ...\n",
      " $ Purchasing_power                             : int  3 4 4 4 3 3 5 3 3 7 ...\n",
      " $ private_third_party_insurance                : int  0 2 2 0 0 0 0 0 0 2 ...\n",
      " $ third_party_insurance_firms                  : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ third_party_insurane_agriculture             : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ car_policies                                 : int  6 0 6 6 0 6 6 0 5 0 ...\n",
      " $ delivery_van_policies                        : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Contribution_motorcycle.scooter              : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ lorry_policies                               : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ trailer_policies                             : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ tractor_policies                             : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ agricultural_machines_policies               : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ moped_policies                               : int  0 0 0 0 0 0 0 3 0 0 ...\n",
      " $ life_insurances                              : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ private_accident_insurance_policies          : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ family_accidents_insurance_policies          : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ disability_insurance_policies                : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ fire_policies                                : int  5 2 2 2 6 0 0 0 0 3 ...\n",
      " $ surfboard_policies                           : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ boat_policies                                : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ bicycle_policies                             : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ property_insurance_policies                  : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ social_security_insurance_policies           : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_private_third_party_insurance      : int  0 2 1 0 0 0 0 0 0 1 ...\n",
      " $ Number_of_third_party_insurance_firms        : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_third_party_insurane_agriculture   : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_car_policies                       : int  1 0 1 1 0 1 1 0 1 0 ...\n",
      " $ Number_of_delivery_van_policies              : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_motorcycle.scooter_policies        : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_lorry_policies                     : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_trailer_policies                   : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_tractor_policies                   : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_agricultural_machines_policies     : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_moped_policies                     : int  0 0 0 0 0 0 0 1 0 0 ...\n",
      " $ Number_of_life_insurances                    : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_private_accident_insurance_policies: int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_family_accidents_insurance_policies: int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_disability_insurance_policies      : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_fire_policies                      : int  1 1 1 1 1 0 0 0 0 1 ...\n",
      " $ Number_of_surfboard_policies                 : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_boat_policies                      : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_bicycle_policies                   : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_property_insurance_policies        : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_social_security_insurance_policies : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_mobile_home_policies               : int  0 0 0 0 0 0 0 0 0 0 ...\n"
     ]
    }
   ],
   "source": [
    "#Check Shape of Data and List Out All Columns\n",
    "str(caravan_df_maindataset)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### 'Number_of_mobile_home_policies' is the binary dependent variable that we will classify in our analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead><tr><th scope=col>Customer_Subtype</th><th scope=col>Number_of_houses</th><th scope=col>Avg_size_household</th><th scope=col>Avg_age</th><th scope=col>Customer_main_type</th><th scope=col>Roman_catholic</th><th scope=col>Protestant</th><th scope=col>Other_religion</th><th scope=col>No_religion</th><th scope=col>Married</th><th scope=col>...</th><th scope=col>Number_of_private_accident_insurance_policies</th><th scope=col>Number_of_family_accidents_insurance_policies</th><th scope=col>Number_of_disability_insurance_policies</th><th scope=col>Number_of_fire_policies</th><th scope=col>Number_of_surfboard_policies</th><th scope=col>Number_of_boat_policies</th><th scope=col>Number_of_bicycle_policies</th><th scope=col>Number_of_property_insurance_policies</th><th scope=col>Number_of_social_security_insurance_policies</th><th scope=col>Number_of_mobile_home_policies</th></tr></thead>\n",
       "<tbody>\n",
       "\t<tr><td>33 </td><td>1  </td><td>3  </td><td>2  </td><td> 8 </td><td>0  </td><td>5  </td><td>1  </td><td>3  </td><td>7  </td><td>...</td><td>0  </td><td>0  </td><td>0  </td><td>1  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td></tr>\n",
       "\t<tr><td>37 </td><td>1  </td><td>2  </td><td>2  </td><td> 8 </td><td>1  </td><td>4  </td><td>1  </td><td>4  </td><td>6  </td><td>...</td><td>0  </td><td>0  </td><td>0  </td><td>1  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td></tr>\n",
       "\t<tr><td>37 </td><td>1  </td><td>2  </td><td>2  </td><td> 8 </td><td>0  </td><td>4  </td><td>2  </td><td>4  </td><td>3  </td><td>...</td><td>0  </td><td>0  </td><td>0  </td><td>1  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td></tr>\n",
       "\t<tr><td> 9 </td><td>1  </td><td>3  </td><td>3  </td><td> 3 </td><td>2  </td><td>3  </td><td>2  </td><td>4  </td><td>5  </td><td>...</td><td>0  </td><td>0  </td><td>0  </td><td>1  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td></tr>\n",
       "\t<tr><td>40 </td><td>1  </td><td>4  </td><td>2  </td><td>10 </td><td>1  </td><td>4  </td><td>1  </td><td>4  </td><td>7  </td><td>...</td><td>0  </td><td>0  </td><td>0  </td><td>1  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td></tr>\n",
       "\t<tr><td>23 </td><td>1  </td><td>2  </td><td>1  </td><td> 5 </td><td>0  </td><td>5  </td><td>0  </td><td>5  </td><td>0  </td><td>...</td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td><td>0  </td></tr>\n",
       "</tbody>\n",
       "</table>\n"
      ],
      "text/latex": [
       "\\begin{tabular}{r|llllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllll}\n",
       " Customer\\_Subtype & Number\\_of\\_houses & Avg\\_size\\_household & Avg\\_age & Customer\\_main\\_type & Roman\\_catholic & Protestant & Other\\_religion & No\\_religion & Married & ... & Number\\_of\\_private\\_accident\\_insurance\\_policies & Number\\_of\\_family\\_accidents\\_insurance\\_policies & Number\\_of\\_disability\\_insurance\\_policies & Number\\_of\\_fire\\_policies & Number\\_of\\_surfboard\\_policies & Number\\_of\\_boat\\_policies & Number\\_of\\_bicycle\\_policies & Number\\_of\\_property\\_insurance\\_policies & Number\\_of\\_social\\_security\\_insurance\\_policies & Number\\_of\\_mobile\\_home\\_policies\\\\\n",
       "\\hline\n",
       "\t 33  & 1   & 3   & 2   &  8  & 0   & 5   & 1   & 3   & 7   & ... & 0   & 0   & 0   & 1   & 0   & 0   & 0   & 0   & 0   & 0  \\\\\n",
       "\t 37  & 1   & 2   & 2   &  8  & 1   & 4   & 1   & 4   & 6   & ... & 0   & 0   & 0   & 1   & 0   & 0   & 0   & 0   & 0   & 0  \\\\\n",
       "\t 37  & 1   & 2   & 2   &  8  & 0   & 4   & 2   & 4   & 3   & ... & 0   & 0   & 0   & 1   & 0   & 0   & 0   & 0   & 0   & 0  \\\\\n",
       "\t  9  & 1   & 3   & 3   &  3  & 2   & 3   & 2   & 4   & 5   & ... & 0   & 0   & 0   & 1   & 0   & 0   & 0   & 0   & 0   & 0  \\\\\n",
       "\t 40  & 1   & 4   & 2   & 10  & 1   & 4   & 1   & 4   & 7   & ... & 0   & 0   & 0   & 1   & 0   & 0   & 0   & 0   & 0   & 0  \\\\\n",
       "\t 23  & 1   & 2   & 1   &  5  & 0   & 5   & 0   & 5   & 0   & ... & 0   & 0   & 0   & 0   & 0   & 0   & 0   & 0   & 0   & 0  \\\\\n",
       "\\end{tabular}\n"
      ],
      "text/markdown": [
       "\n",
       "Customer_Subtype | Number_of_houses | Avg_size_household | Avg_age | Customer_main_type | Roman_catholic | Protestant | Other_religion | No_religion | Married | ... | Number_of_private_accident_insurance_policies | Number_of_family_accidents_insurance_policies | Number_of_disability_insurance_policies | Number_of_fire_policies | Number_of_surfboard_policies | Number_of_boat_policies | Number_of_bicycle_policies | Number_of_property_insurance_policies | Number_of_social_security_insurance_policies | Number_of_mobile_home_policies | \n",
       "|---|---|---|---|---|---|\n",
       "| 33  | 1   | 3   | 2   |  8  | 0   | 5   | 1   | 3   | 7   | ... | 0   | 0   | 0   | 1   | 0   | 0   | 0   | 0   | 0   | 0   | \n",
       "| 37  | 1   | 2   | 2   |  8  | 1   | 4   | 1   | 4   | 6   | ... | 0   | 0   | 0   | 1   | 0   | 0   | 0   | 0   | 0   | 0   | \n",
       "| 37  | 1   | 2   | 2   |  8  | 0   | 4   | 2   | 4   | 3   | ... | 0   | 0   | 0   | 1   | 0   | 0   | 0   | 0   | 0   | 0   | \n",
       "|  9  | 1   | 3   | 3   |  3  | 2   | 3   | 2   | 4   | 5   | ... | 0   | 0   | 0   | 1   | 0   | 0   | 0   | 0   | 0   | 0   | \n",
       "| 40  | 1   | 4   | 2   | 10  | 1   | 4   | 1   | 4   | 7   | ... | 0   | 0   | 0   | 1   | 0   | 0   | 0   | 0   | 0   | 0   | \n",
       "| 23  | 1   | 2   | 1   |  5  | 0   | 5   | 0   | 5   | 0   | ... | 0   | 0   | 0   | 0   | 0   | 0   | 0   | 0   | 0   | 0   | \n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "  Customer_Subtype Number_of_houses Avg_size_household Avg_age\n",
       "1 33               1                3                  2      \n",
       "2 37               1                2                  2      \n",
       "3 37               1                2                  2      \n",
       "4  9               1                3                  3      \n",
       "5 40               1                4                  2      \n",
       "6 23               1                2                  1      \n",
       "  Customer_main_type Roman_catholic Protestant Other_religion No_religion\n",
       "1  8                 0              5          1              3          \n",
       "2  8                 1              4          1              4          \n",
       "3  8                 0              4          2              4          \n",
       "4  3                 2              3          2              4          \n",
       "5 10                 1              4          1              4          \n",
       "6  5                 0              5          0              5          \n",
       "  Married ... Number_of_private_accident_insurance_policies\n",
       "1 7       ... 0                                            \n",
       "2 6       ... 0                                            \n",
       "3 3       ... 0                                            \n",
       "4 5       ... 0                                            \n",
       "5 7       ... 0                                            \n",
       "6 0       ... 0                                            \n",
       "  Number_of_family_accidents_insurance_policies\n",
       "1 0                                            \n",
       "2 0                                            \n",
       "3 0                                            \n",
       "4 0                                            \n",
       "5 0                                            \n",
       "6 0                                            \n",
       "  Number_of_disability_insurance_policies Number_of_fire_policies\n",
       "1 0                                       1                      \n",
       "2 0                                       1                      \n",
       "3 0                                       1                      \n",
       "4 0                                       1                      \n",
       "5 0                                       1                      \n",
       "6 0                                       0                      \n",
       "  Number_of_surfboard_policies Number_of_boat_policies\n",
       "1 0                            0                      \n",
       "2 0                            0                      \n",
       "3 0                            0                      \n",
       "4 0                            0                      \n",
       "5 0                            0                      \n",
       "6 0                            0                      \n",
       "  Number_of_bicycle_policies Number_of_property_insurance_policies\n",
       "1 0                          0                                    \n",
       "2 0                          0                                    \n",
       "3 0                          0                                    \n",
       "4 0                          0                                    \n",
       "5 0                          0                                    \n",
       "6 0                          0                                    \n",
       "  Number_of_social_security_insurance_policies Number_of_mobile_home_policies\n",
       "1 0                                            0                             \n",
       "2 0                                            0                             \n",
       "3 0                                            0                             \n",
       "4 0                                            0                             \n",
       "5 0                                            0                             \n",
       "6 0                                            0                             "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Sample of the Data\n",
    "head(caravan_df_maindataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "'Missing values: 0'"
      ],
      "text/latex": [
       "'Missing values: 0'"
      ],
      "text/markdown": [
       "'Missing values: 0'"
      ],
      "text/plain": [
       "[1] \"Missing values: 0\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Ensure there are no missing values.\n",
    "paste0(\"Missing values: \", sum(is.na(caravan_df_maindataset)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "'data.frame':\t9822 obs. of  86 variables:\n",
      " $ Customer_Subtype                             : int  33 37 37 9 40 23 39 33 33 11 ...\n",
      " $ Number_of_houses                             : int  1 1 1 1 1 1 2 1 1 2 ...\n",
      " $ Avg_size_household                           : int  3 2 2 3 4 2 3 2 2 3 ...\n",
      " $ Avg_age                                      : int  2 2 2 3 2 1 2 3 4 3 ...\n",
      " $ Customer_main_type                           : int  8 8 8 3 10 5 9 8 8 3 ...\n",
      " $ Roman_catholic                               : int  0 1 0 2 1 0 2 0 0 3 ...\n",
      " $ Protestant                                   : int  5 4 4 3 4 5 2 7 1 5 ...\n",
      " $ Other_religion                               : int  1 1 2 2 1 0 0 0 3 0 ...\n",
      " $ No_religion                                  : int  3 4 4 4 4 5 5 2 6 2 ...\n",
      " $ Married                                      : int  7 6 3 5 7 0 7 7 6 7 ...\n",
      " $ Living_together                              : int  0 2 2 2 1 6 2 2 0 0 ...\n",
      " $ Other_relation                               : int  2 2 4 2 2 3 0 0 3 2 ...\n",
      " $ Singles                                      : int  1 0 4 2 2 3 0 0 3 2 ...\n",
      " $ Household_without_children                   : int  2 4 4 3 4 5 3 5 3 2 ...\n",
      " $ Household_with_children                      : int  6 5 2 4 4 2 6 4 3 6 ...\n",
      " $ High_level_education                         : int  1 0 0 3 5 0 0 0 0 0 ...\n",
      " $ Medium_level_education                       : int  2 5 5 4 4 5 4 3 1 4 ...\n",
      " $ Lower_level_education                        : int  7 4 4 2 0 4 5 6 8 5 ...\n",
      " $ High_status                                  : int  1 0 0 4 0 2 0 2 1 2 ...\n",
      " $ Entrepreneur                                 : int  0 0 0 0 5 0 0 0 1 0 ...\n",
      " $ Farmer                                       : int  1 0 0 0 4 0 0 0 0 0 ...\n",
      " $ Middle_management                            : int  2 5 7 3 0 4 4 2 1 3 ...\n",
      " $ Skilled_labourers                            : int  5 0 0 1 0 2 1 5 8 3 ...\n",
      " $ Unskilled_labourers                          : int  2 4 2 2 0 2 5 2 1 3 ...\n",
      " $ Social_class_A                               : int  1 0 0 3 9 2 0 2 1 1 ...\n",
      " $ Social_class_B1                              : int  1 2 5 2 0 2 1 1 1 2 ...\n",
      " $ Social_class_B2                              : int  2 3 0 1 0 2 4 2 0 1 ...\n",
      " $ Social_class_C                               : int  6 5 4 4 0 4 5 5 8 4 ...\n",
      " $ Social_class_D                               : int  1 0 0 0 0 2 0 2 1 2 ...\n",
      " $ Rented_house                                 : int  1 2 7 5 4 9 6 0 9 0 ...\n",
      " $ Home_owners                                  : int  8 7 2 4 5 0 3 9 0 9 ...\n",
      " $ X1_car                                       : int  8 7 7 9 6 5 8 4 5 6 ...\n",
      " $ X2_cars                                      : int  0 1 0 0 2 3 0 4 2 1 ...\n",
      " $ No_car                                       : int  1 2 2 0 1 3 1 2 3 2 ...\n",
      " $ National_Health_Service                      : int  8 6 9 7 5 9 9 6 7 6 ...\n",
      " $ Private_health_insurance                     : int  1 3 0 2 4 0 0 3 2 3 ...\n",
      " $ Income_._30                                  : int  0 2 4 1 0 5 4 2 7 2 ...\n",
      " $ Income_30.45.000                             : int  4 0 5 5 0 2 3 5 2 3 ...\n",
      " $ Income_45.75.000                             : int  5 5 0 3 9 3 3 3 1 3 ...\n",
      " $ Income_75.122.000                            : int  0 2 0 0 0 0 0 0 0 1 ...\n",
      " $ Income_.123.000                              : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Average_income                               : int  4 5 3 4 6 3 3 3 2 4 ...\n",
      " $ Purchasing_power                             : int  3 4 4 4 3 3 5 3 3 7 ...\n",
      " $ private_third_party_insurance                : int  0 2 2 0 0 0 0 0 0 2 ...\n",
      " $ third_party_insurance_firms                  : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ third_party_insurane_agriculture             : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ car_policies                                 : int  6 0 6 6 0 6 6 0 5 0 ...\n",
      " $ delivery_van_policies                        : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Contribution_motorcycle.scooter              : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ lorry_policies                               : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ trailer_policies                             : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ tractor_policies                             : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ agricultural_machines_policies               : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ moped_policies                               : int  0 0 0 0 0 0 0 3 0 0 ...\n",
      " $ life_insurances                              : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ private_accident_insurance_policies          : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ family_accidents_insurance_policies          : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ disability_insurance_policies                : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ fire_policies                                : int  5 2 2 2 6 0 0 0 0 3 ...\n",
      " $ surfboard_policies                           : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ boat_policies                                : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ bicycle_policies                             : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ property_insurance_policies                  : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ social_security_insurance_policies           : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_private_third_party_insurance      : int  0 2 1 0 0 0 0 0 0 1 ...\n",
      " $ Number_of_third_party_insurance_firms        : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_third_party_insurane_agriculture   : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_car_policies                       : int  1 0 1 1 0 1 1 0 1 0 ...\n",
      " $ Number_of_delivery_van_policies              : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_motorcycle.scooter_policies        : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_lorry_policies                     : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_trailer_policies                   : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_tractor_policies                   : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_agricultural_machines_policies     : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_moped_policies                     : int  0 0 0 0 0 0 0 1 0 0 ...\n",
      " $ Number_of_life_insurances                    : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_private_accident_insurance_policies: int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_family_accidents_insurance_policies: int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_disability_insurance_policies      : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_fire_policies                      : int  1 1 1 1 1 0 0 0 0 1 ...\n",
      " $ Number_of_surfboard_policies                 : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_boat_policies                      : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_bicycle_policies                   : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_property_insurance_policies        : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_social_security_insurance_policies : int  0 0 0 0 0 0 0 0 0 0 ...\n",
      " $ Number_of_mobile_home_policies               : int  0 0 0 0 0 0 0 0 0 0 ...\n"
     ]
    }
   ],
   "source": [
    "# Finding which features are numeric.\n",
    "numeric_caravan <- which(sapply(caravan_df_maindataset,is.numeric))\n",
    "str(caravan_df_maindataset[,numeric_caravan])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "###### All features are numeric. Since, Number_of_mobile_home_policies is our dependent variable, let's make sure that it is binary by converting it to a factor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "caravan_df_maindataset$Number_of_mobile_home_policies <- factor(caravan_df_maindataset$Number_of_mobile_home_policies, labels=c(0,1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Let's Check How Unbalanced Our Data Is By Counting The Number of 1s and 0s In The Feature Number_of_mobile_home_policies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {},
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAMAAADKOT/pAAAAP1BMVEUAAAAAv8QzMzNNTU1o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\nAQH/DzGynz75PnABAAAAAElFTkSuQmCC",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "library(ggplot2)\n",
    "library(plyr)\n",
    "\n",
    "\n",
    "Not_Insured_with_caravan  <- sum(caravan_df_maindataset$Number_of_mobile_home_policies == 0)\n",
    "Insured_with_caravan <- sum(caravan_df_maindataset$Number_of_mobile_home_policies == 1)\n",
    "\n",
    "dat <- data.frame(\n",
    "  Policy_status = factor(c(\"Not_Insured_with_caravan\",\"Insured_with_caravan\"), levels=c(\"Not_Insured_with_caravan\",\"Insured_with_caravan\")),\n",
    "  Count = c( Not_Insured_with_caravan , Insured_with_caravan)\n",
    ")\n",
    "\n",
    "ggplot(data=dat, aes(x=Policy_status, y=Count, fill=Policy_status)) +\n",
    "  geom_bar(colour=\"black\", stat=\"identity\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead><tr><th scope=col>Number_of_mobile_home_policies</th><th scope=col>Count</th></tr></thead>\n",
       "<tbody>\n",
       "\t<tr><td>0   </td><td>9236</td></tr>\n",
       "\t<tr><td>1   </td><td> 586</td></tr>\n",
       "</tbody>\n",
       "</table>\n"
      ],
      "text/latex": [
       "\\begin{tabular}{r|ll}\n",
       " Number\\_of\\_mobile\\_home\\_policies & Count\\\\\n",
       "\\hline\n",
       "\t 0    & 9236\\\\\n",
       "\t 1    &  586\\\\\n",
       "\\end{tabular}\n"
      ],
      "text/markdown": [
       "\n",
       "Number_of_mobile_home_policies | Count | \n",
       "|---|---|\n",
       "| 0    | 9236 | \n",
       "| 1    |  586 | \n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "  Number_of_mobile_home_policies Count\n",
       "1 0                              9236 \n",
       "2 1                               586 "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "Frequency.Number_of_mobile_home_policies <- data.frame(Number_of_mobile_home_policies = levels(caravan_df_maindataset$Number_of_mobile_home_policies), Count = as.numeric(table(caravan_df_maindataset$Number_of_mobile_home_policies)))\n",
    "Frequency.Number_of_mobile_home_policies"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### It is clear that our dataset is highly unbalanced with only 5.9% of observations actually buying the insurance. Before we run our analysis and handle the unbalanced data, let's understand the characteristics of the observations that actually bought the 'mobile home insurance'. Since, it is a cross-selling problem, we will focus more on understanding which existing customers tend to also buy the 'mobile_home_insurance'. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 313,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {},
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAMAAADKOT/pAAAASFBMVEUAAAAAsPYAv30zMzNN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\nkKSA/h/X3BVSXacapAAAAABJRU5ErkJggg==",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "TrainDataset <- read.csv('C:\\\\Users\\\\Jayant\\\\Desktop\\\\DMPA\\\\Project\\\\Data_sets\\\\Data_sets\\\\Master_dataset.csv')\n",
    "\n",
    "car_policies <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$car_policies  != 0)\n",
    "Income_30K <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$Income_._30  != 0)\n",
    "moped_policies <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$moped_policies  != 0)\n",
    "fire_policies <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$fire_policies  != 0)\n",
    "Lower_level_education <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$Lower_level_education  != 0)\n",
    "\n",
    "\n",
    "dat <- data.frame(\n",
    "  Selected_Features = factor(c(\"car_policies\" , \"Income_30K \" , \"moped_policies\" , \"fire_policies\"  , \"Lower_level_education\" ), levels=c(\"car_policies\" , \"Income_30K \" , \"moped_policies\" , \"fire_policies\"  , \"Lower_level_education\")),\n",
    "  Count = c( car_policies  ,  Income_30K  , moped_policies , fire_policies  , Lower_level_education )\n",
    ")\n",
    "\n",
    "ggplot(data=dat, aes(x=Selected_Features, y=Count, fill=Selected_Features)) +\n",
    "  geom_bar(colour=\"black\", stat=\"identity\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### The above visualization shows characteristics of only the observations that actually bought the 'mobile home insurance'. It can be observed that existing customers of car_policies and fire_policies tend to also buy the 'mobile home insurance'. Moreover, the success class population is generally lower_level_education and Income_30K. Let's look at more features of the success class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 315,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {},
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAMAAADKOT/pAAAAS1BMVEUAAAAAujgAv8QzMzNN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",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "No_of_boat_policies <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$Number_of_boat_policies != 0)\n",
    "Married <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$Married  != 0)\n",
    "Other_relation <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$Other_relation  != 0)\n",
    "boat_policies <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$boat_policies  != 0)\n",
    "Skilled_labourers <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$Skilled_labourers  != 0)\n",
    "\n",
    "Lower_level_education <- sum(TrainDataset$Number_of_mobile_home_policies == 1 & TrainDataset$Lower_level_education  != 0)\n",
    "\n",
    "\n",
    "dat <- data.frame(\n",
    "  Selected_Features = factor(c(\"No_of_boat_policies\",\"Married\" , \"Other_relation\" , \"boat_policies\" , \"Skilled_labourers\"  , \"Lower_level_education\" ), levels=c(\"No_of_boat_policies\",\"Married\" , \"Other_relation\" , \"boat_policies\" , \"Skilled_labourers\"  , \"Lower_level_education\")),\n",
    "  Count = c(No_of_boat_policies, Married , Other_relation , boat_policies , Skilled_labourers, Lower_level_education)\n",
    ")\n",
    "\n",
    "ggplot(data=dat, aes(x=Selected_Features, y=Count, fill=Selected_Features)) +\n",
    "  geom_bar(colour=\"black\", stat=\"identity\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "###### Both the above plots are based on the features that were indicated as important predictors from our analysis of bagging and random forest."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### 2. Association Rule Mining For Understanding Cross Selling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Since, Caravan had released this data to understand how to cross-sell its 'Mobile Home Policies' to its existing customers, one of the first approaches we took was to devise association rules to understand customers of which existing policies tend to also buy the 'mobile home policy' ."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Apriori\n",
      "\n",
      "Parameter specification:\n",
      " confidence minval smax arem  aval originalSupport maxtime support minlen\n",
      "        0.7    0.1    1 none FALSE            TRUE       5   0.001      1\n",
      " maxlen target   ext\n",
      "     10  rules FALSE\n",
      "\n",
      "Algorithmic control:\n",
      " filter tree heap memopt load sort verbose\n",
      "    0.1 TRUE TRUE  FALSE TRUE    2    TRUE\n",
      "\n",
      "Absolute minimum support count: 5 \n",
      "\n",
      "set item appearances ...[1 item(s)] done [0.00s].\n",
      "set transactions ...[630 item(s), 5822 transaction(s)] done [0.03s].\n",
      "sorting and recoding items ... [542 item(s)] done [0.02s].\n",
      "creating transaction tree ... done [0.01s].\n",
      "checking subsets of size 1 2 3 4"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message in apriori(data = train, parameter = list(supp = 0.001, conf = 0.7), :\n",
      "\"Mining stopped (time limit reached). Only patterns up to a length of 4 returned!\""
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " done [61.08s].\n",
      "writing ... [82 rule(s)] done [4.60s].\n",
      "creating S4 object  ... done [0.86s].\n",
      "     lhs                                                  rhs                                    support confidence     lift\n",
      "[1]  {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_car_policies=1,                                                                                             \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  1.0000000 16.72989\n",
      "[2]  {Avg_age=3,                                                                                                            \n",
      "      Social_class_B2=3,                                                                                                    \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  1.0000000 16.72989\n",
      "[3]  {Social_class_B2=3,                                                                                                    \n",
      "      life_insurances=0,                                                                                                    \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.8750000 14.63865\n",
      "[4]  {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_life_insurances=0,                                                                                          \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.8750000 14.63865\n",
      "[5]  {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_private_third_party_insurance=0,                                                                            \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.8571429 14.33990\n",
      "[6]  {Social_class_B2=3,                                                                                                    \n",
      "      private_third_party_insurance=0,                                                                                      \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.8571429 14.33990\n",
      "[7]  {Number_of_houses=1,                                                                                                   \n",
      "      Social_class_B2=3,                                                                                                    \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.8571429 14.33990\n",
      "[8]  {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[9]  {Social_class_B2=3,                                                                                                    \n",
      "      moped_policies=0,                                                                                                     \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[10] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_moped_policies=0,                                                                                           \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[11] {Social_class_B2=3,                                                                                                    \n",
      "      Contribution_motorcycle.scooter=0,                                                                                    \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[12] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_motorcycle.scooter_policies=0,                                                                              \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[13] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_boat_policies=1,                                                                                            \n",
      "      Number_of_bicycle_policies=0}                    => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[14] {Social_class_B2=3,                                                                                                    \n",
      "      bicycle_policies=0,                                                                                                   \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[15] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_tractor_policies=0,                                                                                         \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[16] {Social_class_B2=3,                                                                                                    \n",
      "      tractor_policies=0,                                                                                                   \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[17] {Social_class_B2=3,                                                                                                    \n",
      "      third_party_insurance_firms=0,                                                                                        \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[18] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_boat_policies=1,                                                                                            \n",
      "      Number_of_social_security_insurance_policies=0}  => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[19] {Social_class_B2=3,                                                                                                    \n",
      "      social_security_insurance_policies=0,                                                                                 \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[20] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_third_party_insurance_firms=0,                                                                              \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[21] {Social_class_B2=3,                                                                                                    \n",
      "      trailer_policies=0,                                                                                                   \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[22] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_trailer_policies=0,                                                                                         \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[23] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_delivery_van_policies=0,                                                                                    \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[24] {Social_class_B2=3,                                                                                                    \n",
      "      delivery_van_policies=0,                                                                                              \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[25] {Social_class_B2=3,                                                                                                    \n",
      "      property_insurance_policies=0,                                                                                        \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[26] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_boat_policies=1,                                                                                            \n",
      "      Number_of_property_insurance_policies=0}         => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[27] {Social_class_B2=3,                                                                                                    \n",
      "      family_accidents_insurance_policies=0,                                                                                \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[28] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_family_accidents_insurance_policies=0,                                                                      \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[29] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_private_accident_insurance_policies=0,                                                                      \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[30] {Social_class_B2=3,                                                                                                    \n",
      "      private_accident_insurance_policies=0,                                                                                \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[31] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_disability_insurance_policies=0,                                                                            \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[32] {Social_class_B2=3,                                                                                                    \n",
      "      disability_insurance_policies=0,                                                                                      \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[33] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_agricultural_machines_policies=0,                                                                           \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[34] {Social_class_B2=3,                                                                                                    \n",
      "      agricultural_machines_policies=0,                                                                                     \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[35] {Social_class_B2=3,                                                                                                    \n",
      "      lorry_policies=0,                                                                                                     \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[36] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_lorry_policies=0,                                                                                           \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[37] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_surfboard_policies=0,                                                                                       \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[38] {Social_class_B2=3,                                                                                                    \n",
      "      surfboard_policies=0,                                                                                                 \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[39] {Middle_management=3,                                                                                                  \n",
      "      Number_of_tractor_policies=0,                                                                                         \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[40] {Middle_management=3,                                                                                                  \n",
      "      tractor_policies=0,                                                                                                   \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[41] {Other_religion=2,                                                                                                     \n",
      "      fire_policies=4,                                                                                                      \n",
      "      Number_of_social_security_insurance_policies=1}  => {Number_of_mobile_home_policies=1} 0.001202336  0.7777778 13.01213\n",
      "[42] {Skilled_labourers=3,                                                                                                  \n",
      "      Number_of_private_third_party_insurance=0,                                                                            \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[43] {Skilled_labourers=3,                                                                                                  \n",
      "      private_third_party_insurance=0,                                                                                      \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[44] {Social_class_B2=3,                                                                                                    \n",
      "      third_party_insurane_agriculture=0,                                                                                   \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[45] {Social_class_B2=3,                                                                                                    \n",
      "      Number_of_third_party_insurane_agriculture=0,                                                                         \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[46] {Avg_age=3,                                                                                                            \n",
      "      Middle_management=3,                                                                                                  \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[47] {Social_class_B1=1,                                                                                                    \n",
      "      car_policies=6,                                                                                                       \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[48] {Social_class_B1=1,                                                                                                    \n",
      "      Number_of_car_policies=1,                                                                                             \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[49] {Avg_age=3,                                                                                                            \n",
      "      Other_relation=2,                                                                                                     \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[50] {Other_relation=2,                                                                                                     \n",
      "      Number_of_private_third_party_insurance=0,                                                                            \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[51] {Other_relation=2,                                                                                                     \n",
      "      private_third_party_insurance=0,                                                                                      \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[52] {Living_together=1,                                                                                                    \n",
      "      car_policies=6,                                                                                                       \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001030574  0.7500000 12.54741\n",
      "[53] {car_policies=6,                                                                                                       \n",
      "      Number_of_private_third_party_insurance=0,                                                                            \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001545861  0.7500000 12.54741\n",
      "[54] {private_third_party_insurance=0,                                                                                      \n",
      "      car_policies=6,                                                                                                       \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001545861  0.7500000 12.54741\n",
      "[55] {Middle_management=3,                                                                                                  \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[56] {Skilled_labourers=3,                                                                                                  \n",
      "      Number_of_car_policies=1,                                                                                             \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[57] {Avg_age=3,                                                                                                            \n",
      "      Skilled_labourers=3,                                                                                                  \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[58] {Number_of_houses=1,                                                                                                   \n",
      "      Middle_management=3,                                                                                                  \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[59] {Middle_management=3,                                                                                                  \n",
      "      moped_policies=0,                                                                                                     \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[60] {Middle_management=3,                                                                                                  \n",
      "      Number_of_moped_policies=0,                                                                                           \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[61] {Middle_management=3,                                                                                                  \n",
      "      life_insurances=0,                                                                                                    \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[62] {Middle_management=3,                                                                                                  \n",
      "      Number_of_life_insurances=0,                                                                                          \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[63] {Middle_management=3,                                                                                                  \n",
      "      Contribution_motorcycle.scooter=0,                                                                                    \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[64] {Middle_management=3,                                                                                                  \n",
      "      Number_of_motorcycle.scooter_policies=0,                                                                              \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[65] {Middle_management=3,                                                                                                  \n",
      "      Number_of_boat_policies=1,                                                                                            \n",
      "      Number_of_bicycle_policies=0}                    => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[66] {Middle_management=3,                                                                                                  \n",
      "      bicycle_policies=0,                                                                                                   \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[67] {Middle_management=3,                                                                                                  \n",
      "      third_party_insurane_agriculture=0,                                                                                   \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[68] {Middle_management=3,                                                                                                  \n",
      "      Number_of_third_party_insurane_agriculture=0,                                                                         \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[69] {Middle_management=3,                                                                                                  \n",
      "      third_party_insurance_firms=0,                                                                                        \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[70] {Middle_management=3,                                                                                                  \n",
      "      Number_of_boat_policies=1,                                                                                            \n",
      "      Number_of_social_security_insurance_policies=0}  => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[71] {Middle_management=3,                                                                                                  \n",
      "      social_security_insurance_policies=0,                                                                                 \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[72] {Middle_management=3,                                                                                                  \n",
      "      Number_of_third_party_insurance_firms=0,                                                                              \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[73] {Middle_management=3,                                                                                                  \n",
      "      Number_of_delivery_van_policies=0,                                                                                    \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[74] {Middle_management=3,                                                                                                  \n",
      "      delivery_van_policies=0,                                                                                              \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[75] {Middle_management=3,                                                                                                  \n",
      "      family_accidents_insurance_policies=0,                                                                                \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[76] {Middle_management=3,                                                                                                  \n",
      "      Number_of_family_accidents_insurance_policies=0,                                                                      \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[77] {Middle_management=3,                                                                                                  \n",
      "      Number_of_disability_insurance_policies=0,                                                                            \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[78] {Middle_management=3,                                                                                                  \n",
      "      disability_insurance_policies=0,                                                                                      \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[79] {Middle_management=3,                                                                                                  \n",
      "      Number_of_agricultural_machines_policies=0,                                                                           \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[80] {Middle_management=3,                                                                                                  \n",
      "      agricultural_machines_policies=0,                                                                                     \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[81] {Middle_management=3,                                                                                                  \n",
      "      lorry_policies=0,                                                                                                     \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n",
      "[82] {Middle_management=3,                                                                                                  \n",
      "      Number_of_lorry_policies=0,                                                                                           \n",
      "      Number_of_boat_policies=1}                       => {Number_of_mobile_home_policies=1} 0.001202336  0.7000000 11.71092\n"
     ]
    }
   ],
   "source": [
    "#Association Rule Mining\n",
    "library(arules)\n",
    "library(arulesViz)\n",
    "\n",
    "train <- read.csv('C:\\\\Users\\\\Jayant\\\\Desktop\\\\DMPA\\\\Project\\\\Data_sets\\\\Data_sets\\\\Train.csv', colClasses=c(rep('factor' , 86)))\n",
    "\n",
    "## Association rule generation\n",
    "rules <- apriori(data = train , parameter = list( supp = 0.001 , conf = 0.7) , appearance = list(default = \"lhs\" , rhs = \"Number_of_mobile_home_policies=1\") )\n",
    "\n",
    "## Sorting rules based on confidence\n",
    "rules <- sort(rules , decreasing = TRUE , by = 'confidence')\n",
    "\n",
    "## Printing the rules\n",
    "inspect(rules)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### As seen above, there exist rules with very high confidence, however, the support of these rules is very less. Since these metrics are data-dependent, lets look at the plot of confidence vs support to adjust them accordingly"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAMAAADKOT/pAAABXFBMVEUAAABNTU1oaGh8fHyM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\nAIQEBEBIQACEBARASEAA/weZ3OCEPB++2QAAAABJRU5ErkJggg==",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "## Ploting the generated rules\n",
    "plot(rules)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### As seen in the above plot, if we take a support of greater than 0.0012, we would not get a rule with a confidence higher than 0.8. This is because our dataset is highly unbalanced and there are only few observations out of almost 6000 training observations that contribute to these rules. Having said that, the best rules we can obtain from this data will have a support of around 0.0012 and a confidence as high as 1."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Apriori\n",
      "\n",
      "Parameter specification:\n",
      " confidence minval smax arem  aval originalSupport maxtime support minlen\n",
      "        0.9    0.1    1 none FALSE            TRUE       5  0.0012      1\n",
      " maxlen target   ext\n",
      "     10  rules FALSE\n",
      "\n",
      "Algorithmic control:\n",
      " filter tree heap memopt load sort verbose\n",
      "    0.1 TRUE TRUE  FALSE TRUE    2    TRUE\n",
      "\n",
      "Absolute minimum support count: 6 \n",
      "\n",
      "set item appearances ...[1 item(s)] done [0.00s].\n",
      "set transactions ...[630 item(s), 5822 transaction(s)] done [0.04s].\n",
      "sorting and recoding items ... [533 item(s)] done [0.01s].\n",
      "creating transaction tree ... done [0.01s].\n",
      "checking subsets of size 1 2 3 4"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message in apriori(data = train, parameter = list(supp = 0.0012, conf = 0.9), :\n",
      "\"Mining stopped (time limit reached). Only patterns up to a length of 4 returned!\""
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " done [66.13s].\n",
      "writing ... [1 rule(s)] done [4.60s].\n",
      "creating S4 object  ... done [0.97s].\n",
      "    lhs                            rhs                                    support confidence     lift\n",
      "[1] {Avg_age=3,                                                                                      \n",
      "     Social_class_B2=3,                                                                              \n",
      "     Number_of_boat_policies=1} => {Number_of_mobile_home_policies=1} 0.001202336          1 16.72989\n"
     ]
    }
   ],
   "source": [
    "#Re-running association rules with support = 0.0012 and confidence = 0.9\n",
    "## Association rule generation\n",
    "rules <- apriori(data = train , parameter = list( supp = 0.0012 , conf = 0.9) , appearance = list(default = \"lhs\" , rhs = \"Number_of_mobile_home_policies=1\") )\n",
    "\n",
    "## Sorting rules based on confidence\n",
    "rules <- sort(rules , decreasing = TRUE , by = 'confidence')\n",
    "\n",
    "## Printing the rules\n",
    "inspect(rules)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### The above rule is associated with the best tradeoff we could get given our data. This rule has a very confidence and hence, can be considered a reliable rule. However, we must keep in mind that the support of this rule is low and it might just occur due to a chance factor.\n",
    "###### This rule indicates that the Caravan's existing customers of boat policies also tend to have Caravan's mobile home policies. This might be a useful insight for Caravan to implement cross selling. However, since this rule has a low support, we decided to dig deep into each of the characteristics of the observations that buy caravan's mobile home policies. We have built classification models that are aimed at understanding who will buy the caravan's mobile home insurance and who will not. Since our dataset is highly imbalanced, we have run each of our classification model using techniques such as oversampling and undersampling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### 3. Classification using Logistic Regression - Unbalanced Data, Undersampled Data, Oversampled Data - Including Business Cost For Each Model "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Logistic Using Unbalanced Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "probabilityDF <- data.frame(matrix(ncol = 20, nrow = 4000))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "#LR Pre-Processing\n",
    "caravan_df_trainLR <- caravan_df_train\n",
    "caravan_df_testLR <- caravan_df_test\n",
    "caravan_df_trainLR$Number_of_mobile_home_policies <- factor(caravan_df_trainLR$Number_of_mobile_home_policies, labels = c(0,1))\n",
    "library(dummies)\n",
    "caravan_df_trainLR <- dummy.data.frame(caravan_df_trainLR, sep = \".\", names = c(\"Customer_main_type\",\"Customer_Subtype\"))\n",
    "caravan_df_testLR <- dummy.data.frame(caravan_df_testLR, sep = \".\", names = c(\"Customer_main_type\",\"Customer_Subtype\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message:\n",
      "\"glm.fit: fitted probabilities numerically 0 or 1 occurred\""
     ]
    },
    {
     "data": {
      "text/plain": [
       "\n",
       "Call:  glm(formula = Number_of_mobile_home_policies ~ ., family = \"binomial\", \n",
       "    data = caravan_df_trainLR)\n",
       "\n",
       "Coefficients:\n",
       "                                  (Intercept)  \n",
       "                                   255.779835  \n",
       "                           Customer_Subtype.1  \n",
       "                                     0.298357  \n",
       "                           Customer_Subtype.2  \n",
       "                                     0.626285  \n",
       "                           Customer_Subtype.3  \n",
       "                                     0.775459  \n",
       "                           Customer_Subtype.4  \n",
       "                                    -0.483366  \n",
       "                           Customer_Subtype.5  \n",
       "                                     0.403773  \n",
       "                           Customer_Subtype.6  \n",
       "                                     0.603530  \n",
       "                           Customer_Subtype.7  \n",
       "                                    -0.218449  \n",
       "                           Customer_Subtype.8  \n",
       "                                     0.752301  \n",
       "                           Customer_Subtype.9  \n",
       "                                     0.336168  \n",
       "                          Customer_Subtype.10  \n",
       "                                    -0.292822  \n",
       "                          Customer_Subtype.11  \n",
       "                                     0.263270  \n",
       "                          Customer_Subtype.12  \n",
       "                                     1.084364  \n",
       "                          Customer_Subtype.13  \n",
       "                                     0.269408  \n",
       "                          Customer_Subtype.15  \n",
       "                                   -13.212560  \n",
       "                          Customer_Subtype.16  \n",
       "                                   -14.385494  \n",
       "                          Customer_Subtype.17  \n",
       "                                   -16.008856  \n",
       "                          Customer_Subtype.18  \n",
       "                                   -14.447273  \n",
       "                          Customer_Subtype.19  \n",
       "                                   -15.059721  \n",
       "                          Customer_Subtype.20  \n",
       "                                     1.240721  \n",
       "                          Customer_Subtype.21  \n",
       "                                   -14.007891  \n",
       "                          Customer_Subtype.22  \n",
       "                                     0.474294  \n",
       "                          Customer_Subtype.23  \n",
       "                                    -0.387975  \n",
       "                          Customer_Subtype.24  \n",
       "                                     0.469392  \n",
       "                          Customer_Subtype.25  \n",
       "                                     0.031868  \n",
       "                          Customer_Subtype.26  \n",
       "                                     0.704459  \n",
       "                          Customer_Subtype.27  \n",
       "                                     0.257646  \n",
       "                          Customer_Subtype.28  \n",
       "                                   -14.331670  \n",
       "                          Customer_Subtype.29  \n",
       "                                     0.088167  \n",
       "                          Customer_Subtype.30  \n",
       "                                     0.799390  \n",
       "                          Customer_Subtype.31  \n",
       "                                     0.603076  \n",
       "                          Customer_Subtype.32  \n",
       "                                     0.971908  \n",
       "                          Customer_Subtype.33  \n",
       "                                     0.754069  \n",
       "                          Customer_Subtype.34  \n",
       "                                     0.425013  \n",
       "                          Customer_Subtype.35  \n",
       "                                     0.224519  \n",
       "                          Customer_Subtype.36  \n",
       "                                     0.927485  \n",
       "                          Customer_Subtype.37  \n",
       "                                     1.036779  \n",
       "                          Customer_Subtype.38  \n",
       "                                     0.940528  \n",
       "                          Customer_Subtype.39  \n",
       "                                     0.600708  \n",
       "                          Customer_Subtype.40  \n",
       "                                   -14.246329  \n",
       "                          Customer_Subtype.41  \n",
       "                                           NA  \n",
       "                             Number_of_houses  \n",
       "                                    -0.079240  \n",
       "                           Avg_size_household  \n",
       "                                    -0.033481  \n",
       "                                      Avg_age  \n",
       "                                     0.190077  \n",
       "                         Customer_main_type.1  \n",
       "                                           NA  \n",
       "                         Customer_main_type.2  \n",
       "                                           NA  \n",
       "                         Customer_main_type.3  \n",
       "                                           NA  \n",
       "                         Customer_main_type.4  \n",
       "                                           NA  \n",
       "                         Customer_main_type.5  \n",
       "                                           NA  \n",
       "                         Customer_main_type.6  \n",
       "                                           NA  \n",
       "                         Customer_main_type.7  \n",
       "                                           NA  \n",
       "                         Customer_main_type.8  \n",
       "                                           NA  \n",
       "                         Customer_main_type.9  \n",
       "                                           NA  \n",
       "                        Customer_main_type.10  \n",
       "                                           NA  \n",
       "                               Roman_catholic  \n",
       "                                    -0.095381  \n",
       "                                   Protestant  \n",
       "                                     0.008046  \n",
       "                               Other_religion  \n",
       "                                     0.010400  \n",
       "                                  No_religion  \n",
       "                                    -0.048281  \n",
       "                                      Married  \n",
       "                                     0.253760  \n",
       "                              Living_together  \n",
       "                                     0.150091  \n",
       "                               Other_relation  \n",
       "                                     0.173822  \n",
       "                                      Singles  \n",
       "                                    -0.063262  \n",
       "                   Household_without_children  \n",
       "                                    -0.122386  \n",
       "                      Household_with_children  \n",
       "                                    -0.091298  \n",
       "                         High_level_education  \n",
       "                                     0.006776  \n",
       "                       Medium_level_education  \n",
       "                                    -0.105271  \n",
       "                        Lower_level_education  \n",
       "                                    -0.238215  \n",
       "                                  High_status  \n",
       "                                     0.089426  \n",
       "                                 Entrepreneur  \n",
       "                                     0.016853  \n",
       "                                       Farmer  \n",
       "                                    -0.046193  \n",
       "                            Middle_management  \n",
       "                                     0.118589  \n",
       "                            Skilled_labourers  \n",
       "                                     0.032230  \n",
       "                          Unskilled_labourers  \n",
       "                                     0.097689  \n",
       "                               Social_class_A  \n",
       "                                    -0.002523  \n",
       "                              Social_class_B1  \n",
       "                                    -0.018820  \n",
       "                              Social_class_B2  \n",
       "                                     0.019203  \n",
       "                               Social_class_C  \n",
       "                                     0.081814  \n",
       "                               Social_class_D  \n",
       "                                    -0.053007  \n",
       "                                 Rented_house  \n",
       "                                   -14.748144  \n",
       "                                  Home_owners  \n",
       "                                   -14.726939  \n",
       "                                       X1_car  \n",
       "                                     0.151550  \n",
       "                                      X2_cars  \n",
       "                                     0.127871  \n",
       "                                       No_car  \n",
       "                                     0.075850  \n",
       "                      National_Health_Service  \n",
       "                                   -14.617326  \n",
       "                     Private_health_insurance  \n",
       "                                   -14.674899  \n",
       "                                  Income_._30  \n",
       "                                     0.104669  \n",
       "                             Income_30.45.000  \n",
       "                                     0.131024  \n",
       "                             Income_45.75.000  \n",
       "                                     0.105565  \n",
       "                            Income_75.122.000  \n",
       "                                     0.114998  \n",
       "                              Income_.123.000  \n",
       "                                    -0.177658  \n",
       "                               Average_income  \n",
       "                                     0.054570  \n",
       "                             Purchasing_power  \n",
       "                                     0.074645  \n",
       "                private_third_party_insurance  \n",
       "                                     0.694217  \n",
       "                  third_party_insurance_firms  \n",
       "                                    -0.197197  \n",
       "             third_party_insurane_agriculture  \n",
       "                                    -0.608277  \n",
       "                                 car_policies  \n",
       "                                     0.222972  \n",
       "                        delivery_van_policies  \n",
       "                                    12.392454  \n",
       "              Contribution_motorcycle.scooter  \n",
       "                                    -0.127699  \n",
       "                               lorry_policies  \n",
       "                                    -2.033866  \n",
       "                             trailer_policies  \n",
       "                                     1.317891  \n",
       "                             tractor_policies  \n",
       "                                     0.749037  \n",
       "               agricultural_machines_policies  \n",
       "                                    -6.008076  \n",
       "                               moped_policies  \n",
       "                                     0.262304  \n",
       "                              life_insurances  \n",
       "                                    -0.251409  \n",
       "          private_accident_insurance_policies  \n",
       "                                    -0.554243  \n",
       "          family_accidents_insurance_policies  \n",
       "                                     1.510621  \n",
       "                disability_insurance_policies  \n",
       "                                     0.761072  \n",
       "                                fire_policies  \n",
       "                                     0.205090  \n",
       "                           surfboard_policies  \n",
       "                                    -8.470268  \n",
       "                                boat_policies  \n",
       "                                    -0.194273  \n",
       "                             bicycle_policies  \n",
       "                                     0.404949  \n",
       "                  property_insurance_policies  \n",
       "                                    -1.025409  \n",
       "           social_security_insurance_policies  \n",
       "                                    -0.163979  \n",
       "      Number_of_private_third_party_insurance  \n",
       "                                    -1.092999  \n",
       "        Number_of_third_party_insurance_firms  \n",
       "                                     0.287025  \n",
       "   Number_of_third_party_insurane_agriculture  \n",
       "                                     1.019005  \n",
       "                       Number_of_car_policies  \n",
       "                                    -0.009499  \n",
       "              Number_of_delivery_van_policies  \n",
       "                                   -74.316981  \n",
       "        Number_of_motorcycle.scooter_policies  \n",
       "                                     0.390978  \n",
       "                     Number_of_lorry_policies  \n",
       "                                    -5.128840  \n",
       "                   Number_of_trailer_policies  \n",
       "                                    -1.772590  \n",
       "                   Number_of_tractor_policies  \n",
       "                                    -2.497866  \n",
       "     Number_of_agricultural_machines_policies  \n",
       "                                     0.450817  \n",
       "                     Number_of_moped_policies  \n",
       "                                    -1.269353  \n",
       "                    Number_of_life_insurances  \n",
       "                                     0.516435  \n",
       "Number_of_private_accident_insurance_policies  \n",
       "                                     0.606855  \n",
       "Number_of_family_accidents_insurance_policies  \n",
       "                                    -3.326548  \n",
       "      Number_of_disability_insurance_policies  \n",
       "                                    -2.977123  \n",
       "                      Number_of_fire_policies  \n",
       "                                    -0.319787  \n",
       "                 Number_of_surfboard_policies  \n",
       "                                    10.340019  \n",
       "                      Number_of_boat_policies  \n",
       "                                     2.510073  \n",
       "                   Number_of_bicycle_policies  \n",
       "                                     0.167198  \n",
       "        Number_of_property_insurance_policies  \n",
       "                                     1.966078  \n",
       " Number_of_social_security_insurance_policies  \n",
       "                                     1.016541  \n",
       "\n",
       "Degrees of Freedom: 5821 Total (i.e. Null);  5699 Residual\n",
       "Null Deviance:\t    2636 \n",
       "Residual Deviance: 2204 \tAIC: 2450"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Classify Using Logistic Regression\n",
    "logisticTrainingFit <- glm(Number_of_mobile_home_policies ~ ., family = \"binomial\", data = caravan_df_trainLR)\n",
    "logisticTrainingFit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message in predict.lm(object, newdata, se.fit, scale = 1, type = ifelse(type == :\n",
      "\"prediction from a rank-deficient fit may be misleading\""
     ]
    },
    {
     "data": {
      "text/plain": [
       "          LOutPredictedClass\n",
       "LOutActual    0    1\n",
       "         0 3757    5\n",
       "         1  235    3"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Predict Class And Display Confusion Matrix\n",
    "LOutPredicted <- predict(logisticTrainingFit, caravan_df_testLR, type = \"response\")\n",
    "LOutPredictedClass <- ifelse(LOutPredicted>0.5, 1, 0)\n",
    "LOutActual <- caravan_df_testLR$Number_of_mobile_home_policies\n",
    "LConfusionOutPredicted <- table(LOutActual, LOutPredictedClass)\n",
    "rownames(LConfusionOutPredicted) <- c(\"0\",\"1\")\n",
    "colnames(LConfusionOutPredicted) <- c(\"0\",\"1\")\n",
    "LConfusionOutPredicted\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$LR[i] <- LOutPredicted[i]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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\nri3YAAAAAElFTkSuQmCC",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC and AUC for LR\n",
    "probs <- LOutPredicted\n",
    "library(ROCR)\n",
    "LRPred <- prediction(probs, caravan_df_testLR$Number_of_mobile_home_policies)\n",
    "LRPerf <- performance(LRPred, \"tpr\", \"fpr\")\n",
    "plot(LRPerf, colorize=TRUE)\n",
    "abline(a=0, b=1, lty=2, lwd=3, col=\"black\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "An object of class \"performance\"\n",
       "Slot \"x.name\":\n",
       "[1] \"None\"\n",
       "\n",
       "Slot \"y.name\":\n",
       "[1] \"Area under the ROC curve\"\n",
       "\n",
       "Slot \"alpha.name\":\n",
       "[1] \"none\"\n",
       "\n",
       "Slot \"x.values\":\n",
       "list()\n",
       "\n",
       "Slot \"y.values\":\n",
       "[[1]]\n",
       "[1] 0.7209613\n",
       "\n",
       "\n",
       "Slot \"alpha.values\":\n",
       "list()\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#AUC\n",
    "performance(LRPred, \"auc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<dl class=dl-horizontal>\n",
       "\t<dt>acc</dt>\n",
       "\t\t<dd>0.94</dd>\n",
       "\t<dt>sens</dt>\n",
       "\t\t<dd>0.0126050420168067</dd>\n",
       "\t<dt>spec</dt>\n",
       "\t\t<dd>0.998670919723551</dd>\n",
       "\t<dt>ppv</dt>\n",
       "\t\t<dd>0.375</dd>\n",
       "\t<dt>npv</dt>\n",
       "\t\t<dd>0.941132264529058</dd>\n",
       "\t<dt>lor</dt>\n",
       "\t\t<dd>2.26096490766382</dd>\n",
       "</dl>\n"
      ],
      "text/latex": [
       "\\begin{description*}\n",
       "\\item[acc] 0.94\n",
       "\\item[sens] 0.0126050420168067\n",
       "\\item[spec] 0.998670919723551\n",
       "\\item[ppv] 0.375\n",
       "\\item[npv] 0.941132264529058\n",
       "\\item[lor] 2.26096490766382\n",
       "\\end{description*}\n"
      ],
      "text/markdown": [
       "acc\n",
       ":   0.94sens\n",
       ":   0.0126050420168067spec\n",
       ":   0.998670919723551ppv\n",
       ":   0.375npv\n",
       ":   0.941132264529058lor\n",
       ":   2.26096490766382\n",
       "\n"
      ],
      "text/plain": [
       "       acc       sens       spec        ppv        npv        lor \n",
       "0.94000000 0.01260504 0.99867092 0.37500000 0.94113226 2.26096491 \n",
       "attr(,\"negative\")\n",
       "[1] \"Not Purchased\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Corresponding Performance Measures\n",
    "LRPrediction <- factor(as.factor(LOutPredictedClass), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "LRActual <- factor(as.factor(caravan_df_testLR$Number_of_mobile_home_policies), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "library(crossval)\n",
    "CMLR <- confusionMatrix(LRActual, LRPrediction, negative = \"Not Purchased\" )\n",
    "diagnosticErrors(CMLR)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Logistic Regression Using Undersampled Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead><tr><th scope=col>Number_of_mobile_home_policies</th><th scope=col>Count</th></tr></thead>\n",
       "<tbody>\n",
       "\t<tr><td>0  </td><td>348</td></tr>\n",
       "\t<tr><td>1  </td><td>348</td></tr>\n",
       "</tbody>\n",
       "</table>\n"
      ],
      "text/latex": [
       "\\begin{tabular}{r|ll}\n",
       " Number\\_of\\_mobile\\_home\\_policies & Count\\\\\n",
       "\\hline\n",
       "\t 0   & 348\\\\\n",
       "\t 1   & 348\\\\\n",
       "\\end{tabular}\n"
      ],
      "text/markdown": [
       "\n",
       "Number_of_mobile_home_policies | Count | \n",
       "|---|---|\n",
       "| 0   | 348 | \n",
       "| 1   | 348 | \n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "  Number_of_mobile_home_policies Count\n",
       "1 0                              348  \n",
       "2 1                              348  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Under Sampling Data\n",
    "#Taking all the observations with dependent variable = 1\n",
    "caravan_df_train_LR_Under <- caravan_df_trainLR[caravan_df_trainLR$Number_of_mobile_home_policies==1,]\n",
    "\n",
    "#Randomly select observations with dependent variable = 0\n",
    "zeroObs <- caravan_df_trainLR[caravan_df_trainLR$Number_of_mobile_home_policies==0,]\n",
    "set.seed(123457)\n",
    "rearrangedZeroObs <-  zeroObs[sample(nrow(zeroObs), length(caravan_df_train_LR_Under$Number_of_mobile_home_policies)),]\n",
    "\n",
    "#Appending rows of randomly selected 0s in our undersampled data frame\n",
    "caravan_df_train_LR_Under <- rbind(caravan_df_train_LR_Under, rearrangedZeroObs)\n",
    "\n",
    "#Let's verify that number of 1s and 0s in our undersampled data are equal\n",
    "undersampled.Frequency.Number_of_mobile_home_policies <- data.frame(Number_of_mobile_home_policies = levels(as.factor(caravan_df_train_LR_Under$Number_of_mobile_home_policies)), Count = as.numeric(table(caravan_df_train_LR_Under$Number_of_mobile_home_policies)))\n",
    "undersampled.Frequency.Number_of_mobile_home_policies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 222,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message:\n",
      "\"glm.fit: fitted probabilities numerically 0 or 1 occurred\""
     ]
    },
    {
     "data": {
      "text/plain": [
       "\n",
       "Call:  glm(formula = Number_of_mobile_home_policies ~ ., family = \"binomial\", \n",
       "    data = caravan_df_train_LR_Under)\n",
       "\n",
       "Coefficients:\n",
       "                                  (Intercept)  \n",
       "                                    2.845e+02  \n",
       "                           Customer_Subtype.1  \n",
       "                                    1.154e+00  \n",
       "                           Customer_Subtype.2  \n",
       "                                    1.533e+00  \n",
       "                           Customer_Subtype.3  \n",
       "                                    1.563e+00  \n",
       "                           Customer_Subtype.4  \n",
       "                                    8.962e-01  \n",
       "                           Customer_Subtype.5  \n",
       "                                   -6.599e-02  \n",
       "                           Customer_Subtype.6  \n",
       "                                    1.983e+00  \n",
       "                           Customer_Subtype.7  \n",
       "                                    1.793e+01  \n",
       "                           Customer_Subtype.8  \n",
       "                                    1.612e+00  \n",
       "                           Customer_Subtype.9  \n",
       "                                    7.786e-01  \n",
       "                          Customer_Subtype.10  \n",
       "                                    8.128e-01  \n",
       "                          Customer_Subtype.11  \n",
       "                                    1.136e+00  \n",
       "                          Customer_Subtype.12  \n",
       "                                    2.481e+00  \n",
       "                          Customer_Subtype.13  \n",
       "                                    5.320e-01  \n",
       "                          Customer_Subtype.15  \n",
       "                                           NA  \n",
       "                          Customer_Subtype.16  \n",
       "                                   -1.430e+01  \n",
       "                          Customer_Subtype.17  \n",
       "                                           NA  \n",
       "                          Customer_Subtype.18  \n",
       "                                   -1.634e+01  \n",
       "                          Customer_Subtype.19  \n",
       "                                           NA  \n",
       "                          Customer_Subtype.20  \n",
       "                                    1.729e+01  \n",
       "                          Customer_Subtype.21  \n",
       "                                   -1.507e+01  \n",
       "                          Customer_Subtype.22  \n",
       "                                    6.193e-01  \n",
       "                          Customer_Subtype.23  \n",
       "                                    4.956e-02  \n",
       "                          Customer_Subtype.24  \n",
       "                                    4.829e-02  \n",
       "                          Customer_Subtype.25  \n",
       "                                    3.287e+00  \n",
       "                          Customer_Subtype.26  \n",
       "                                    2.479e+00  \n",
       "                          Customer_Subtype.27  \n",
       "                                   -8.775e-01  \n",
       "                          Customer_Subtype.28  \n",
       "                                           NA  \n",
       "                          Customer_Subtype.29  \n",
       "                                    1.145e-01  \n",
       "                          Customer_Subtype.30  \n",
       "                                    6.002e-01  \n",
       "                          Customer_Subtype.31  \n",
       "                                    1.122e+00  \n",
       "                          Customer_Subtype.32  \n",
       "                                    9.893e-01  \n",
       "                          Customer_Subtype.33  \n",
       "                                    8.015e-01  \n",
       "                          Customer_Subtype.34  \n",
       "                                    1.389e-01  \n",
       "                          Customer_Subtype.35  \n",
       "                                    4.939e-01  \n",
       "                          Customer_Subtype.36  \n",
       "                                    1.380e+00  \n",
       "                          Customer_Subtype.37  \n",
       "                                    1.545e+00  \n",
       "                          Customer_Subtype.38  \n",
       "                                    9.188e-01  \n",
       "                          Customer_Subtype.39  \n",
       "                                    9.092e-01  \n",
       "                          Customer_Subtype.40  \n",
       "                                   -1.425e+01  \n",
       "                          Customer_Subtype.41  \n",
       "                                           NA  \n",
       "                             Number_of_houses  \n",
       "                                   -4.991e-01  \n",
       "                           Avg_size_household  \n",
       "                                   -1.013e-01  \n",
       "                                      Avg_age  \n",
       "                                    2.632e-01  \n",
       "                         Customer_main_type.1  \n",
       "                                           NA  \n",
       "                         Customer_main_type.2  \n",
       "                                           NA  \n",
       "                         Customer_main_type.3  \n",
       "                                           NA  \n",
       "                         Customer_main_type.4  \n",
       "                                           NA  \n",
       "                         Customer_main_type.5  \n",
       "                                           NA  \n",
       "                         Customer_main_type.6  \n",
       "                                           NA  \n",
       "                         Customer_main_type.7  \n",
       "                                           NA  \n",
       "                         Customer_main_type.8  \n",
       "                                           NA  \n",
       "                         Customer_main_type.9  \n",
       "                                           NA  \n",
       "                        Customer_main_type.10  \n",
       "                                           NA  \n",
       "                               Roman_catholic  \n",
       "                                   -2.849e-01  \n",
       "                                   Protestant  \n",
       "                                   -9.688e-02  \n",
       "                               Other_religion  \n",
       "                                   -1.309e-01  \n",
       "                                  No_religion  \n",
       "                                   -2.402e-01  \n",
       "                                      Married  \n",
       "                                    1.802e-01  \n",
       "                              Living_together  \n",
       "                                    2.085e-01  \n",
       "                               Other_relation  \n",
       "                                    4.010e-02  \n",
       "                                      Singles  \n",
       "                                    1.417e-01  \n",
       "                   Household_without_children  \n",
       "                                    9.639e-02  \n",
       "                      Household_with_children  \n",
       "                                    1.056e-01  \n",
       "                         High_level_education  \n",
       "                                    5.254e-01  \n",
       "                       Medium_level_education  \n",
       "                                    2.203e-01  \n",
       "                        Lower_level_education  \n",
       "                                    6.661e-02  \n",
       "                                  High_status  \n",
       "                                    9.755e-02  \n",
       "                                 Entrepreneur  \n",
       "                                   -1.624e-01  \n",
       "                                       Farmer  \n",
       "                                   -1.568e-01  \n",
       "                            Middle_management  \n",
       "                                    1.995e-01  \n",
       "                            Skilled_labourers  \n",
       "                                    3.077e-02  \n",
       "                          Unskilled_labourers  \n",
       "                                   -4.224e-03  \n",
       "                               Social_class_A  \n",
       "                                   -2.346e-02  \n",
       "                              Social_class_B1  \n",
       "                                   -4.219e-04  \n",
       "                              Social_class_B2  \n",
       "                                    7.517e-02  \n",
       "                               Social_class_C  \n",
       "                                    2.555e-01  \n",
       "                               Social_class_D  \n",
       "                                    1.964e-01  \n",
       "                                 Rented_house  \n",
       "                                   -1.632e+01  \n",
       "                                  Home_owners  \n",
       "                                   -1.631e+01  \n",
       "                                       X1_car  \n",
       "                                    5.120e-01  \n",
       "                                      X2_cars  \n",
       "                                    4.484e-01  \n",
       "                                       No_car  \n",
       "                                    2.646e-01  \n",
       "                      National_Health_Service  \n",
       "                                   -1.649e+01  \n",
       "                     Private_health_insurance  \n",
       "                                   -1.669e+01  \n",
       "                                  Income_._30  \n",
       "                                   -3.182e-02  \n",
       "                             Income_30.45.000  \n",
       "                                   -3.259e-03  \n",
       "                             Income_45.75.000  \n",
       "                                   -1.252e-01  \n",
       "                            Income_75.122.000  \n",
       "                                   -1.453e-01  \n",
       "                              Income_.123.000  \n",
       "                                   -2.222e-01  \n",
       "                               Average_income  \n",
       "                                    2.684e-01  \n",
       "                             Purchasing_power  \n",
       "                                   -3.437e-02  \n",
       "                private_third_party_insurance  \n",
       "                                    4.143e-01  \n",
       "                  third_party_insurance_firms  \n",
       "                                   -1.299e+00  \n",
       "             third_party_insurane_agriculture  \n",
       "                                   -1.595e+00  \n",
       "                                 car_policies  \n",
       "                                    1.479e-01  \n",
       "                        delivery_van_policies  \n",
       "                                    7.585e+00  \n",
       "              Contribution_motorcycle.scooter  \n",
       "                                   -6.894e-01  \n",
       "                               lorry_policies  \n",
       "                                   -5.666e+00  \n",
       "                             trailer_policies  \n",
       "                                   -2.474e+00  \n",
       "                             tractor_policies  \n",
       "                                    1.133e+00  \n",
       "               agricultural_machines_policies  \n",
       "                                   -5.446e+00  \n",
       "                               moped_policies  \n",
       "                                    4.381e-01  \n",
       "                              life_insurances  \n",
       "                                   -1.204e-01  \n",
       "          private_accident_insurance_policies  \n",
       "                                   -1.127e+01  \n",
       "          family_accidents_insurance_policies  \n",
       "                                   -2.346e-01  \n",
       "                disability_insurance_policies  \n",
       "                                    2.939e+00  \n",
       "                                fire_policies  \n",
       "                                    2.887e-01  \n",
       "                           surfboard_policies  \n",
       "                                   -5.329e-02  \n",
       "                                boat_policies  \n",
       "                                    1.232e-01  \n",
       "                             bicycle_policies  \n",
       "                                   -2.584e-01  \n",
       "                  property_insurance_policies  \n",
       "                                    4.802e-01  \n",
       "           social_security_insurance_policies  \n",
       "                                    3.700e-01  \n",
       "      Number_of_private_third_party_insurance  \n",
       "                                   -7.178e-01  \n",
       "        Number_of_third_party_insurance_firms  \n",
       "                                    3.196e+00  \n",
       "   Number_of_third_party_insurane_agriculture  \n",
       "                                    5.201e+00  \n",
       "                       Number_of_car_policies  \n",
       "                                    4.627e-01  \n",
       "              Number_of_delivery_van_policies  \n",
       "                                   -2.836e+01  \n",
       "        Number_of_motorcycle.scooter_policies  \n",
       "                                    2.794e+00  \n",
       "                     Number_of_lorry_policies  \n",
       "                                           NA  \n",
       "                   Number_of_trailer_policies  \n",
       "                                    4.774e+00  \n",
       "                   Number_of_tractor_policies  \n",
       "                                   -3.827e+00  \n",
       "     Number_of_agricultural_machines_policies  \n",
       "                                           NA  \n",
       "                     Number_of_moped_policies  \n",
       "                                   -2.113e+00  \n",
       "                    Number_of_life_insurances  \n",
       "                                    2.260e-01  \n",
       "Number_of_private_accident_insurance_policies  \n",
       "                                    2.077e+01  \n",
       "Number_of_family_accidents_insurance_policies  \n",
       "                                    3.517e-01  \n",
       "      Number_of_disability_insurance_policies  \n",
       "                                           NA  \n",
       "                      Number_of_fire_policies  \n",
       "                                   -6.915e-01  \n",
       "                 Number_of_surfboard_policies  \n",
       "                                           NA  \n",
       "                      Number_of_boat_policies  \n",
       "                                    2.133e+00  \n",
       "                   Number_of_bicycle_policies  \n",
       "                                    1.153e+00  \n",
       "        Number_of_property_insurance_policies  \n",
       "                                    1.539e+01  \n",
       " Number_of_social_security_insurance_policies  \n",
       "                                   -1.199e-01  \n",
       "\n",
       "Degrees of Freedom: 695 Total (i.e. Null);  581 Residual\n",
       "Null Deviance:\t    964.9 \n",
       "Residual Deviance: 684.9 \tAIC: 914.9"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Classify Using Logistic Regression with Undersampling\n",
    "logisticTrainingFit <- glm(Number_of_mobile_home_policies ~ ., family = \"binomial\", data = caravan_df_train_LR_Under)\n",
    "logisticTrainingFit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 223,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message in predict.lm(object, newdata, se.fit, scale = 1, type = ifelse(type == :\n",
      "\"prediction from a rank-deficient fit may be misleading\""
     ]
    },
    {
     "data": {
      "text/plain": [
       "          LOutPredictedClass\n",
       "LOutActual    0    1\n",
       "         0 2405 1357\n",
       "         1   83  155"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Predict Class And Display Confusion Matrix\n",
    "LOutPredicted <- predict(logisticTrainingFit, caravan_df_testLR, type = \"response\")\n",
    "LOutPredictedClass <- ifelse(LOutPredicted>0.5, 1, 0)\n",
    "LOutActual <- caravan_df_testLR$Number_of_mobile_home_policies\n",
    "LConfusionOutPredicted <- table(LOutActual, LOutPredictedClass)\n",
    "rownames(LConfusionOutPredicted) <- c(\"0\",\"1\")\n",
    "colnames(LConfusionOutPredicted) <- c(\"0\",\"1\")\n",
    "LConfusionOutPredicted\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$LRU[i] <- LOutPredicted[i]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC and AUC for LR\n",
    "probs <- LOutPredicted\n",
    "library(ROCR)\n",
    "LRPred <- prediction(probs, caravan_df_testLR$Number_of_mobile_home_policies)\n",
    "LRPerf <- performance(LRPred, \"tpr\", \"fpr\")\n",
    "plot(LRPerf, colorize=TRUE)\n",
    "abline(a=0, b=1, lty=2, lwd=3, col=\"black\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "An object of class \"performance\"\n",
       "Slot \"x.name\":\n",
       "[1] \"None\"\n",
       "\n",
       "Slot \"y.name\":\n",
       "[1] \"Area under the ROC curve\"\n",
       "\n",
       "Slot \"alpha.name\":\n",
       "[1] \"none\"\n",
       "\n",
       "Slot \"x.values\":\n",
       "list()\n",
       "\n",
       "Slot \"y.values\":\n",
       "[[1]]\n",
       "[1] 0.6820203\n",
       "\n",
       "\n",
       "Slot \"alpha.values\":\n",
       "list()\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#AUC\n",
    "performance(LRPred, \"auc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<dl class=dl-horizontal>\n",
       "\t<dt>acc</dt>\n",
       "\t\t<dd>0.64</dd>\n",
       "\t<dt>sens</dt>\n",
       "\t\t<dd>0.651260504201681</dd>\n",
       "\t<dt>spec</dt>\n",
       "\t\t<dd>0.639287612971823</dd>\n",
       "\t<dt>ppv</dt>\n",
       "\t\t<dd>0.102513227513228</dd>\n",
       "\t<dt>npv</dt>\n",
       "\t\t<dd>0.966639871382637</dd>\n",
       "\t<dt>lor</dt>\n",
       "\t\t<dd>1.19685803182764</dd>\n",
       "</dl>\n"
      ],
      "text/latex": [
       "\\begin{description*}\n",
       "\\item[acc] 0.64\n",
       "\\item[sens] 0.651260504201681\n",
       "\\item[spec] 0.639287612971823\n",
       "\\item[ppv] 0.102513227513228\n",
       "\\item[npv] 0.966639871382637\n",
       "\\item[lor] 1.19685803182764\n",
       "\\end{description*}\n"
      ],
      "text/markdown": [
       "acc\n",
       ":   0.64sens\n",
       ":   0.651260504201681spec\n",
       ":   0.639287612971823ppv\n",
       ":   0.102513227513228npv\n",
       ":   0.966639871382637lor\n",
       ":   1.19685803182764\n",
       "\n"
      ],
      "text/plain": [
       "      acc      sens      spec       ppv       npv       lor \n",
       "0.6400000 0.6512605 0.6392876 0.1025132 0.9666399 1.1968580 \n",
       "attr(,\"negative\")\n",
       "[1] \"Not Purchased\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Corresponding Performance Measures\n",
    "LRPrediction <- factor(as.factor(LOutPredictedClass), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "LRActual <- factor(as.factor(caravan_df_testLR$Number_of_mobile_home_policies), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "library(crossval)\n",
    "CMLR <- confusionMatrix(LRActual, LRPrediction, negative = \"Not Purchased\" )\n",
    "diagnosticErrors(CMLR)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Logistic Regression Using Oversampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "'Ratio of 1s to 0s- 1:15.7298850574713'"
      ],
      "text/latex": [
       "'Ratio of 1s to 0s- 1:15.7298850574713'"
      ],
      "text/markdown": [
       "'Ratio of 1s to 0s- 1:15.7298850574713'"
      ],
      "text/plain": [
       "[1] \"Ratio of 1s to 0s- 1:15.7298850574713\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#First let's recall the ratio of 1s and 0s in our original training dataset\n",
    "paste0(\"Ratio of 1s to 0s- 1:\", as.numeric(table(caravan_df_trainLR$Number_of_mobile_home_policies))[1]/as.numeric(table(caravan_df_trainLR$Number_of_mobile_home_policies))[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead><tr><th scope=col>Number_of_mobile_home_policies</th><th scope=col>Count</th></tr></thead>\n",
       "<tbody>\n",
       "\t<tr><td>0   </td><td>5474</td></tr>\n",
       "\t<tr><td>1   </td><td>2436</td></tr>\n",
       "</tbody>\n",
       "</table>\n"
      ],
      "text/latex": [
       "\\begin{tabular}{r|ll}\n",
       " Number\\_of\\_mobile\\_home\\_policies & Count\\\\\n",
       "\\hline\n",
       "\t 0    & 5474\\\\\n",
       "\t 1    & 2436\\\\\n",
       "\\end{tabular}\n"
      ],
      "text/markdown": [
       "\n",
       "Number_of_mobile_home_policies | Count | \n",
       "|---|---|\n",
       "| 0    | 5474 | \n",
       "| 1    | 2436 | \n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "  Number_of_mobile_home_policies Count\n",
       "1 0                              5474 \n",
       "2 1                              2436 "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Now let's duplicate the observations with dependent variable = 1 to make the ratio approximately 1:2\n",
    "caravan_df_train_LR_OverDummy <- caravan_df_trainLR[caravan_df_trainLR$Number_of_mobile_home_policies==1,]\n",
    "caravan_df_train_LR_Over <- NULL\n",
    "for (i in 1:7){\n",
    "    caravan_df_train_LR_Over <- rbind(caravan_df_train_LR_Over, caravan_df_train_LR_OverDummy)\n",
    "}\n",
    "caravan_df_train_LR_Over <- rbind(caravan_df_train_LR_Over, caravan_df_trainLR[caravan_df_trainLR$Number_of_mobile_home_policies==0,])\n",
    "#Let's verify the number of 1s and 0s in our oversampled data\n",
    "oversampled.Frequency.Number_of_mobile_home_policies <- data.frame(Number_of_mobile_home_policies = levels(as.factor(caravan_df_train_LR_Over$Number_of_mobile_home_policies)), Count = as.numeric(table(caravan_df_train_LR_Over$Number_of_mobile_home_policies)))\n",
    "oversampled.Frequency.Number_of_mobile_home_policies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 226,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "'Ratio of 1s to 0s After Oversampling- 1:2.24712643678161'"
      ],
      "text/latex": [
       "'Ratio of 1s to 0s After Oversampling- 1:2.24712643678161'"
      ],
      "text/markdown": [
       "'Ratio of 1s to 0s After Oversampling- 1:2.24712643678161'"
      ],
      "text/plain": [
       "[1] \"Ratio of 1s to 0s After Oversampling- 1:2.24712643678161\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Let's verify the ratio of 1s to 0s after over sampling\n",
    "paste0(\"Ratio of 1s to 0s After Oversampling- 1:\", as.numeric(table(caravan_df_train_LR_Over$Number_of_mobile_home_policies))[1]/as.numeric(table(caravan_df_train_LR_Over$Number_of_mobile_home_policies))[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 227,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message:\n",
      "\"glm.fit: fitted probabilities numerically 0 or 1 occurred\""
     ]
    },
    {
     "data": {
      "text/plain": [
       "\n",
       "Call:  glm(formula = Number_of_mobile_home_policies ~ ., family = \"binomial\", \n",
       "    data = caravan_df_train_LR_Over)\n",
       "\n",
       "Coefficients:\n",
       "                                  (Intercept)  \n",
       "                                   271.983944  \n",
       "                           Customer_Subtype.1  \n",
       "                                     0.188433  \n",
       "                           Customer_Subtype.2  \n",
       "                                     0.471638  \n",
       "                           Customer_Subtype.3  \n",
       "                                     0.782514  \n",
       "                           Customer_Subtype.4  \n",
       "                                    -0.327808  \n",
       "                           Customer_Subtype.5  \n",
       "                                     0.657871  \n",
       "                           Customer_Subtype.6  \n",
       "                                     0.363745  \n",
       "                           Customer_Subtype.7  \n",
       "                                     0.029915  \n",
       "                           Customer_Subtype.8  \n",
       "                                     0.808550  \n",
       "                           Customer_Subtype.9  \n",
       "                                     0.440991  \n",
       "                          Customer_Subtype.10  \n",
       "                                    -0.334054  \n",
       "                          Customer_Subtype.11  \n",
       "                                     0.370786  \n",
       "                          Customer_Subtype.12  \n",
       "                                     0.933006  \n",
       "                          Customer_Subtype.13  \n",
       "                                     0.256710  \n",
       "                          Customer_Subtype.15  \n",
       "                                   -13.674826  \n",
       "                          Customer_Subtype.16  \n",
       "                                   -14.939199  \n",
       "                          Customer_Subtype.17  \n",
       "                                   -16.964669  \n",
       "                          Customer_Subtype.18  \n",
       "                                   -15.062066  \n",
       "                          Customer_Subtype.19  \n",
       "                                   -15.916252  \n",
       "                          Customer_Subtype.20  \n",
       "                                     1.080937  \n",
       "                          Customer_Subtype.21  \n",
       "                                   -14.779102  \n",
       "                          Customer_Subtype.22  \n",
       "                                     0.716716  \n",
       "                          Customer_Subtype.23  \n",
       "                                    -0.343226  \n",
       "                          Customer_Subtype.24  \n",
       "                                     0.356492  \n",
       "                          Customer_Subtype.25  \n",
       "                                     0.312662  \n",
       "                          Customer_Subtype.26  \n",
       "                                     0.947308  \n",
       "                          Customer_Subtype.27  \n",
       "                                     0.243666  \n",
       "                          Customer_Subtype.28  \n",
       "                                   -14.926236  \n",
       "                          Customer_Subtype.29  \n",
       "                                     0.159922  \n",
       "                          Customer_Subtype.30  \n",
       "                                     0.864254  \n",
       "                          Customer_Subtype.31  \n",
       "                                     0.953216  \n",
       "                          Customer_Subtype.32  \n",
       "                                     1.285804  \n",
       "                          Customer_Subtype.33  \n",
       "                                     0.859242  \n",
       "                          Customer_Subtype.34  \n",
       "                                     0.406664  \n",
       "                          Customer_Subtype.35  \n",
       "                                     0.045833  \n",
       "                          Customer_Subtype.36  \n",
       "                                     1.194953  \n",
       "                          Customer_Subtype.37  \n",
       "                                     1.047054  \n",
       "                          Customer_Subtype.38  \n",
       "                                     1.087663  \n",
       "                          Customer_Subtype.39  \n",
       "                                     0.676084  \n",
       "                          Customer_Subtype.40  \n",
       "                                   -14.904161  \n",
       "                          Customer_Subtype.41  \n",
       "                                           NA  \n",
       "                             Number_of_houses  \n",
       "                                    -0.180631  \n",
       "                           Avg_size_household  \n",
       "                                    -0.029992  \n",
       "                                      Avg_age  \n",
       "                                     0.193919  \n",
       "                         Customer_main_type.1  \n",
       "                                           NA  \n",
       "                         Customer_main_type.2  \n",
       "                                           NA  \n",
       "                         Customer_main_type.3  \n",
       "                                           NA  \n",
       "                         Customer_main_type.4  \n",
       "                                           NA  \n",
       "                         Customer_main_type.5  \n",
       "                                           NA  \n",
       "                         Customer_main_type.6  \n",
       "                                           NA  \n",
       "                         Customer_main_type.7  \n",
       "                                           NA  \n",
       "                         Customer_main_type.8  \n",
       "                                           NA  \n",
       "                         Customer_main_type.9  \n",
       "                                           NA  \n",
       "                        Customer_main_type.10  \n",
       "                                           NA  \n",
       "                               Roman_catholic  \n",
       "                                    -0.047251  \n",
       "                                   Protestant  \n",
       "                                     0.056095  \n",
       "                               Other_religion  \n",
       "                                     0.067429  \n",
       "                                  No_religion  \n",
       "                                    -0.002044  \n",
       "                                      Married  \n",
       "                                     0.241859  \n",
       "                              Living_together  \n",
       "                                     0.170279  \n",
       "                               Other_relation  \n",
       "                                     0.133098  \n",
       "                                      Singles  \n",
       "                                    -0.030952  \n",
       "                   Household_without_children  \n",
       "                                    -0.117416  \n",
       "                      Household_with_children  \n",
       "                                    -0.079265  \n",
       "                         High_level_education  \n",
       "                                     0.022175  \n",
       "                       Medium_level_education  \n",
       "                                    -0.084494  \n",
       "                        Lower_level_education  \n",
       "                                    -0.222939  \n",
       "                                  High_status  \n",
       "                                     0.122539  \n",
       "                                 Entrepreneur  \n",
       "                                     0.046054  \n",
       "                                       Farmer  \n",
       "                                    -0.042243  \n",
       "                            Middle_management  \n",
       "                                     0.116425  \n",
       "                            Skilled_labourers  \n",
       "                                     0.022335  \n",
       "                          Unskilled_labourers  \n",
       "                                     0.085407  \n",
       "                               Social_class_A  \n",
       "                                    -0.005715  \n",
       "                              Social_class_B1  \n",
       "                                    -0.027602  \n",
       "                              Social_class_B2  \n",
       "                                     0.008061  \n",
       "                               Social_class_C  \n",
       "                                     0.132007  \n",
       "                               Social_class_D  \n",
       "                                    -0.052836  \n",
       "                                 Rented_house  \n",
       "                                   -15.623346  \n",
       "                                  Home_owners  \n",
       "                                   -15.605815  \n",
       "                                       X1_car  \n",
       "                                     0.201651  \n",
       "                                      X2_cars  \n",
       "                                     0.136822  \n",
       "                                       No_car  \n",
       "                                     0.123143  \n",
       "                      National_Health_Service  \n",
       "                                   -15.447528  \n",
       "                     Private_health_insurance  \n",
       "                                   -15.498660  \n",
       "                                  Income_._30  \n",
       "                                     0.072746  \n",
       "                             Income_30.45.000  \n",
       "                                     0.100949  \n",
       "                             Income_45.75.000  \n",
       "                                     0.059630  \n",
       "                            Income_75.122.000  \n",
       "                                     0.070916  \n",
       "                              Income_.123.000  \n",
       "                                    -0.248372  \n",
       "                               Average_income  \n",
       "                                     0.090805  \n",
       "                             Purchasing_power  \n",
       "                                     0.112506  \n",
       "                private_third_party_insurance  \n",
       "                                     0.606498  \n",
       "                  third_party_insurance_firms  \n",
       "                                    -0.323572  \n",
       "             third_party_insurane_agriculture  \n",
       "                                    -0.468062  \n",
       "                                 car_policies  \n",
       "                                     0.201491  \n",
       "                        delivery_van_policies  \n",
       "                                    13.574296  \n",
       "              Contribution_motorcycle.scooter  \n",
       "                                    -0.141111  \n",
       "                               lorry_policies  \n",
       "                                    -2.082615  \n",
       "                             trailer_policies  \n",
       "                                     1.146329  \n",
       "                             tractor_policies  \n",
       "                                     0.741338  \n",
       "               agricultural_machines_policies  \n",
       "                                    -4.938477  \n",
       "                               moped_policies  \n",
       "                                     0.184188  \n",
       "                              life_insurances  \n",
       "                                    -0.213984  \n",
       "          private_accident_insurance_policies  \n",
       "                                    -0.625009  \n",
       "          family_accidents_insurance_policies  \n",
       "                                     1.067929  \n",
       "                disability_insurance_policies  \n",
       "                                     1.252419  \n",
       "                                fire_policies  \n",
       "                                     0.278646  \n",
       "                           surfboard_policies  \n",
       "                                    -8.078434  \n",
       "                                boat_policies  \n",
       "                                    -0.132402  \n",
       "                             bicycle_policies  \n",
       "                                     0.476115  \n",
       "                  property_insurance_policies  \n",
       "                                    -0.946242  \n",
       "           social_security_insurance_policies  \n",
       "                                    -0.144246  \n",
       "      Number_of_private_third_party_insurance  \n",
       "                                    -0.924381  \n",
       "        Number_of_third_party_insurance_firms  \n",
       "                                     0.586345  \n",
       "   Number_of_third_party_insurane_agriculture  \n",
       "                                     0.300524  \n",
       "                       Number_of_car_policies  \n",
       "                                     0.079586  \n",
       "              Number_of_delivery_van_policies  \n",
       "                                   -80.995850  \n",
       "        Number_of_motorcycle.scooter_policies  \n",
       "                                     0.339577  \n",
       "                     Number_of_lorry_policies  \n",
       "                                    -6.022607  \n",
       "                   Number_of_trailer_policies  \n",
       "                                    -1.430297  \n",
       "                   Number_of_tractor_policies  \n",
       "                                    -2.456457  \n",
       "     Number_of_agricultural_machines_policies  \n",
       "                                    -2.355871  \n",
       "                     Number_of_moped_policies  \n",
       "                                    -1.301124  \n",
       "                    Number_of_life_insurances  \n",
       "                                     0.405107  \n",
       "Number_of_private_accident_insurance_policies  \n",
       "                                     0.395737  \n",
       "Number_of_family_accidents_insurance_policies  \n",
       "                                    -2.246126  \n",
       "      Number_of_disability_insurance_policies  \n",
       "                                    -5.624873  \n",
       "                      Number_of_fire_policies  \n",
       "                                    -0.565400  \n",
       "                 Number_of_surfboard_policies  \n",
       "                                     8.529587  \n",
       "                      Number_of_boat_policies  \n",
       "                                     3.208989  \n",
       "                   Number_of_bicycle_policies  \n",
       "                                     0.156314  \n",
       "        Number_of_property_insurance_policies  \n",
       "                                     1.634689  \n",
       " Number_of_social_security_insurance_policies  \n",
       "                                     1.047642  \n",
       "\n",
       "Degrees of Freedom: 7909 Total (i.e. Null);  7787 Residual\n",
       "Null Deviance:\t    9768 \n",
       "Residual Deviance: 7627 \tAIC: 7873"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Classify Using Logistic Regression with Oversampling\n",
    "logisticTrainingFit <- glm(Number_of_mobile_home_policies ~ ., family = \"binomial\", data = caravan_df_train_LR_Over)\n",
    "logisticTrainingFit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message in predict.lm(object, newdata, se.fit, scale = 1, type = ifelse(type == :\n",
      "\"prediction from a rank-deficient fit may be misleading\""
     ]
    },
    {
     "data": {
      "text/plain": [
       "          LOutPredictedClass\n",
       "LOutActual    0    1\n",
       "         0 3348  414\n",
       "         1  154   84"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Predict Class And Display Confusion Matrix\n",
    "LOutPredicted <- predict(logisticTrainingFit, caravan_df_testLR, type = \"response\")\n",
    "LOutPredictedClass <- ifelse(LOutPredicted>0.5, 1, 0)\n",
    "LOutActual <- caravan_df_testLR$Number_of_mobile_home_policies\n",
    "LConfusionOutPredicted <- table(LOutActual, LOutPredictedClass)\n",
    "rownames(LConfusionOutPredicted) <- c(\"0\",\"1\")\n",
    "colnames(LConfusionOutPredicted) <- c(\"0\",\"1\")\n",
    "LConfusionOutPredicted\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$LRO[i] <- LOutPredicted[i]\n",
    "    }\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$Actual[i] <- LOutActual[i]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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\nAElFTkSuQmCC",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC and AUC for LR\n",
    "probs <- LOutPredicted\n",
    "library(ROCR)\n",
    "LRPred <- prediction(probs, caravan_df_testLR$Number_of_mobile_home_policies)\n",
    "LRPerf <- performance(LRPred, \"tpr\", \"fpr\")\n",
    "plot(LRPerf, colorize=TRUE)\n",
    "abline(a=0, b=1, lty=2, lwd=3, col=\"black\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "An object of class \"performance\"\n",
       "Slot \"x.name\":\n",
       "[1] \"None\"\n",
       "\n",
       "Slot \"y.name\":\n",
       "[1] \"Area under the ROC curve\"\n",
       "\n",
       "Slot \"alpha.name\":\n",
       "[1] \"none\"\n",
       "\n",
       "Slot \"x.values\":\n",
       "list()\n",
       "\n",
       "Slot \"y.values\":\n",
       "[[1]]\n",
       "[1] 0.7209925\n",
       "\n",
       "\n",
       "Slot \"alpha.values\":\n",
       "list()\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#AUC\n",
    "performance(LRPred, \"auc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<dl class=dl-horizontal>\n",
       "\t<dt>acc</dt>\n",
       "\t\t<dd>0.858</dd>\n",
       "\t<dt>sens</dt>\n",
       "\t\t<dd>0.352941176470588</dd>\n",
       "\t<dt>spec</dt>\n",
       "\t\t<dd>0.889952153110048</dd>\n",
       "\t<dt>ppv</dt>\n",
       "\t\t<dd>0.168674698795181</dd>\n",
       "\t<dt>npv</dt>\n",
       "\t\t<dd>0.956025128498001</dd>\n",
       "\t<dt>lor</dt>\n",
       "\t\t<dd>1.48411665421373</dd>\n",
       "</dl>\n"
      ],
      "text/latex": [
       "\\begin{description*}\n",
       "\\item[acc] 0.858\n",
       "\\item[sens] 0.352941176470588\n",
       "\\item[spec] 0.889952153110048\n",
       "\\item[ppv] 0.168674698795181\n",
       "\\item[npv] 0.956025128498001\n",
       "\\item[lor] 1.48411665421373\n",
       "\\end{description*}\n"
      ],
      "text/markdown": [
       "acc\n",
       ":   0.858sens\n",
       ":   0.352941176470588spec\n",
       ":   0.889952153110048ppv\n",
       ":   0.168674698795181npv\n",
       ":   0.956025128498001lor\n",
       ":   1.48411665421373\n",
       "\n"
      ],
      "text/plain": [
       "      acc      sens      spec       ppv       npv       lor \n",
       "0.8580000 0.3529412 0.8899522 0.1686747 0.9560251 1.4841167 \n",
       "attr(,\"negative\")\n",
       "[1] \"Not Purchased\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Corresponding Performance Measures\n",
    "LRPrediction <- factor(as.factor(LOutPredictedClass), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "LRActual <- factor(as.factor(caravan_df_testLR$Number_of_mobile_home_policies), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "library(crossval)\n",
    "CMLR <- confusionMatrix(LRActual, LRPrediction, negative = \"Not Purchased\" )\n",
    "diagnosticErrors(CMLR)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### 4. Classification using KNN - Unbalanced Data, Undersampled Data, Oversampled Data - Including Business Cost For Each Model "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### KNN Using Unbalanced Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#Classify Using KNN\n",
    "#Normalizing each variable\n",
    "fun <- function(x)\n",
    "{    \n",
    "    a <- mean(x)    \n",
    "    b <- sd(x)    \n",
    "    (x - a)/(b)\n",
    "}\n",
    "library(class)\n",
    "\n",
    "caravan_df_train_normalized <- apply(caravan_df_train, 2, fun)\n",
    "caravan_df_test_normalized <- apply(caravan_df_test, 2, fun)\n",
    "\n",
    "train_input <- as.matrix(caravan_df_train_normalized[,-86])\n",
    "train_output <- as.vector(caravan_df_train_normalized[,86])\n",
    "test_input <- as.matrix(caravan_df_test_normalized[,-86])\n",
    "\n",
    "kmax <- 15\n",
    "ER1 <- rep(0,kmax)\n",
    "ER2 <- rep(0,kmax)\n",
    "\n",
    "for (i in 1:kmax){\n",
    "    prediction <- knn(train_input, train_input,train_output, k=i)\n",
    "    prediction2 <- knn(train_input, test_input, train_output, k=i)\n",
    "\n",
    "#Confusion Matrix For Test Data\n",
    "CM2 <- table(prediction2, caravan_df_test_normalized[,'Number_of_mobile_home_policies'])\n",
    "\n",
    "#Error Rate on the test sample\n",
    "ER2[i] <- (CM2[1,2]+CM2[2,1])/sum(CM2)\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 301,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAMAAADKOT/pAAAAM1BMVEUAAAAAAP9NTU1oaGh8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\nCAkC/gOMS9qWs5h6ewAAAABJRU5ErkJggg==",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Plotting the %Error on the test sample\n",
    "plot(c(1,kmax),c(0,0.2),type=\"n\", xlab=\"k\",ylab=\"Error Rate\")\n",
    "lines(ER2,col=\"blue\")\n",
    "legend(12, 0.2, c(\"Test\"),lty=c(1,1), col=c(\"blue\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 302,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Minimum Validation Error k: 14"
     ]
    }
   ],
   "source": [
    "# Minimum Validation Error k\n",
    "z <- which.min(ER2)\n",
    "cat(\"Minimum Validation Error k:\", z)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 303,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "      predicted\n",
       "actual    0    1\n",
       "     0 3758    4\n",
       "     1  237    1"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Running knn for Minimum Validation Error k\n",
    "prediction2 <- knn(train_input, test_input, train_output, k=z)\n",
    "actual <- ifelse(caravan_df_test_normalized[,'Number_of_mobile_home_policies']>0,1,0)\n",
    "predicted <- ifelse(as.numeric(prediction2)==1,0,1)\n",
    "table(actual, predicted)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 307,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Running knn for Minimum Validation Error k with prob=TRUE\n",
    "prediction2 <- knn(train_input, test_input, train_output, k=15, prob = TRUE)\n",
    "prob <- attr(prediction2, \"prob\")\n",
    "# Plot ROC and AUC for KNN\n",
    "library(ROCR)\n",
    "KNNPred <- prediction(prob, caravan_df_test_normalized[,'Number_of_mobile_home_policies'])\n",
    "KNNPerf <- performance(KNNPred, \"tpr\", \"fpr\")\n",
    "plot(KNNPerf, colorize=TRUE)\n",
    "abline(a=0, b=1, lty=2, lwd=3, col=\"black\")\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$KNN[i] <- prob[i]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "An object of class \"performance\"\n",
       "Slot \"x.name\":\n",
       "[1] \"None\"\n",
       "\n",
       "Slot \"y.name\":\n",
       "[1] \"Area under the ROC curve\"\n",
       "\n",
       "Slot \"alpha.name\":\n",
       "[1] \"none\"\n",
       "\n",
       "Slot \"x.values\":\n",
       "list()\n",
       "\n",
       "Slot \"y.values\":\n",
       "[[1]]\n",
       "[1] 0.3485306\n",
       "\n",
       "\n",
       "Slot \"alpha.values\":\n",
       "list()\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#AUC\n",
    "performance(KNNPred, \"auc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<dl class=dl-horizontal>\n",
       "\t<dt>acc</dt>\n",
       "\t\t<dd>0.94</dd>\n",
       "\t<dt>sens</dt>\n",
       "\t\t<dd>0</dd>\n",
       "\t<dt>spec</dt>\n",
       "\t\t<dd>0.999468367889421</dd>\n",
       "\t<dt>ppv</dt>\n",
       "\t\t<dd>0</dd>\n",
       "\t<dt>npv</dt>\n",
       "\t\t<dd>0.940470235117559</dd>\n",
       "\t<dt>lor</dt>\n",
       "\t\t<dd>-Inf</dd>\n",
       "</dl>\n"
      ],
      "text/latex": [
       "\\begin{description*}\n",
       "\\item[acc] 0.94\n",
       "\\item[sens] 0\n",
       "\\item[spec] 0.999468367889421\n",
       "\\item[ppv] 0\n",
       "\\item[npv] 0.940470235117559\n",
       "\\item[lor] -Inf\n",
       "\\end{description*}\n"
      ],
      "text/markdown": [
       "acc\n",
       ":   0.94sens\n",
       ":   0spec\n",
       ":   0.999468367889421ppv\n",
       ":   0npv\n",
       ":   0.940470235117559lor\n",
       ":   -Inf\n",
       "\n"
      ],
      "text/plain": [
       "      acc      sens      spec       ppv       npv       lor \n",
       "0.9400000 0.0000000 0.9994684 0.0000000 0.9404702      -Inf \n",
       "attr(,\"negative\")\n",
       "[1] \"Not Purchased\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Corresponding Performance Measures\n",
    "KNNPrediction <- factor(as.factor(prediction2), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "KNNActual <- factor(as.factor(caravan_df_test_normalized[,'Number_of_mobile_home_policies']), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "library(crossval)\n",
    "CMNB <- confusionMatrix(KNNActual, KNNPrediction, negative = \"Not Purchased\" )\n",
    "diagnosticErrors(CMNB)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### KNN Using Undersampled Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead><tr><th scope=col>Number_of_mobile_home_policies</th><th scope=col>Count</th></tr></thead>\n",
       "<tbody>\n",
       "\t<tr><td>0  </td><td>348</td></tr>\n",
       "\t<tr><td>1  </td><td>348</td></tr>\n",
       "</tbody>\n",
       "</table>\n"
      ],
      "text/latex": [
       "\\begin{tabular}{r|ll}\n",
       " Number\\_of\\_mobile\\_home\\_policies & Count\\\\\n",
       "\\hline\n",
       "\t 0   & 348\\\\\n",
       "\t 1   & 348\\\\\n",
       "\\end{tabular}\n"
      ],
      "text/markdown": [
       "\n",
       "Number_of_mobile_home_policies | Count | \n",
       "|---|---|\n",
       "| 0   | 348 | \n",
       "| 1   | 348 | \n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "  Number_of_mobile_home_policies Count\n",
       "1 0                              348  \n",
       "2 1                              348  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Under Sampling Data\n",
    "#Taking all the observations with dependent variable = 1\n",
    "caravan_df_train_KNN_Under <- caravan_df_train[caravan_df_train$Number_of_mobile_home_policies==1,]\n",
    "\n",
    "#Randomly select observations with dependent variable = 0\n",
    "zeroObs <- caravan_df_train[caravan_df_train$Number_of_mobile_home_policies==0,]\n",
    "set.seed(123457)\n",
    "rearrangedZeroObs <-  zeroObs[sample(nrow(zeroObs), length(caravan_df_train_KNN_Under$Customer_Subtype)),]\n",
    "\n",
    "#Appending rows of randomly selected 0s in our undersampled data frame\n",
    "caravan_df_train_KNN_Under <- rbind(caravan_df_train_KNN_Under, rearrangedZeroObs)\n",
    "\n",
    "#Let's verify that number of 1s and 0s in our undersampled data are equal\n",
    "undersampled.Frequency.Number_of_mobile_home_policies <- data.frame(Number_of_mobile_home_policies = levels(as.factor(caravan_df_train_KNN_Under$Number_of_mobile_home_policies)), Count = as.numeric(table(caravan_df_train_KNN_Under$Number_of_mobile_home_policies)))\n",
    "undersampled.Frequency.Number_of_mobile_home_policies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 250,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAMAAADKOT/pAAAAM1BMVEUAAAAAAP9NTU1oaGh8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\nAAAAAElFTkSuQmCC",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Classify Using KNN\n",
    "#Normalizing each variable\n",
    "fun <- function(x)\n",
    "{    \n",
    "    a <- mean(x)    \n",
    "    b <- sd(x)    \n",
    "    (x - a)/(b)\n",
    "}\n",
    "library(class)\n",
    "\n",
    "caravan_df_train_normalized <- apply(caravan_df_train_KNN_Under, 2, fun)\n",
    "caravan_df_test_normalized <- apply(caravan_df_test, 2, fun)\n",
    "\n",
    "train_input <- as.matrix(caravan_df_train_normalized[,-86])\n",
    "train_output <- as.vector(caravan_df_train_normalized[,86])\n",
    "test_input <- as.matrix(caravan_df_test_normalized[,-86])\n",
    "\n",
    "kmax <- 15\n",
    "ER1 <- rep(0,kmax)\n",
    "ER2 <- rep(0,kmax)\n",
    "\n",
    "for (i in 1:kmax){\n",
    "    prediction <- knn(train_input, train_input,train_output, k=i)\n",
    "    prediction2 <- knn(train_input, test_input, train_output, k=i)\n",
    "\n",
    "#Confusion Matrix For Test Data\n",
    "CM2 <- table(prediction2, caravan_df_test_normalized[,'Number_of_mobile_home_policies'])\n",
    "\n",
    "#Error Rate on the test sample\n",
    "ER2[i] <- (CM2[1,2]+CM2[2,1])/sum(CM2)\n",
    "}\n",
    "#Plotting the %Error on the test sample\n",
    "plot(c(1,kmax),c(0.15,0.8),type=\"n\", xlab=\"k\",ylab=\"Error Rate\")\n",
    "lines(ER2, col=\"blue\")\n",
    "legend(12, 0.75, c(\"Test\"),lty=c(1,1), col=c(\"blue\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Minimum Validation Error k: 11"
     ]
    }
   ],
   "source": [
    "# Minimum Validation Error k\n",
    "z <- which.min(ER2)\n",
    "cat(\"Minimum Validation Error k:\", z)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 252,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "      predicted\n",
       "actual    0    1\n",
       "     0 2104 1658\n",
       "     1   73  165"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Running knn for Minimum Validation Error k\n",
    "prediction2 <- knn(train_input, test_input, train_output, k=z)\n",
    "actual <- ifelse(caravan_df_test_normalized[,'Number_of_mobile_home_policies']>0,1,0)\n",
    "predicted <- ifelse(as.numeric(prediction2)==1,0,1)\n",
    "table(actual, predicted)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 253,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94\nAAAgAElEQVR4nO3df4ijeZ4f9q/sCeQIoca5dU922ul24Kj2YUO3J2Y96zHXTndgPXdWncE7\nl+42w/3hmbaaFPiOLkjvuCpmXIVvCCoz+JZMUT0Y4oGRmLFNpoS3c+aqjgncVdvEc6Uc66PF\nBlIiw6TEnlMy2cOEjZU/nl2tRlKpVJKeR8/z6PWi/6h69Ej61tPVj979+f4qdDqdAABA9v2R\neTcAAIDZEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJC\nsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7\nAICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMA\nyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICc\nEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnB\nDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwA\nAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAg\nJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJC\nsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7\nAICcEOwAAHJCsAMAyAnBDgAgJwQ7AICceG7eDciGQqEw7yYAACnS6XTm3YQhBLtxpfPvDwBI\nTPoLPbpiAQDOl/5UFwQ7AIBz9aa6NHfiCXYAAKNkJdUFwQ4AYIQ0pLqDg4OVlZVCofDgwYOD\ng4MRZwp2AADDJZ/qms3m9vZ2oVBYWVmJMlytVrt9+3atVgsh7Ozs3L59e2Nj48wGp7yimBKF\nggsFAIsoynZ9MSCmYNBsNq9evdp7pFKp3L17d3d395d+6ZeWlpZardYHH3ywtra2v79/69at\nwVdQsQMAOFOn00msuPMbv/EbIYRKpRK9aZTqisXim2++ubS0FEK4dOnSG2+8EUL4+OOPh76C\nQtRYVOwAgK6YgsFgdbBQKJTL5YcPH44+rStLCxQ3Go1nz541Go21tbXe4+Vy+fLlyy+99NLy\n8vK82gYAMHOHh4d/+Id/OHi8WCwOPT8bwa7dbj969GhnZ2foo92ct76+vrm5mWC7AABmZnd3\n9/79+48fP45G1IUQXn755d4T2u32Rx99FEK4d+/e0FfIRg9jtVqN+pjfeuutF1988cqVK72P\ntlqtL774YmNjo1arVSqVO3fuzLwBumIBgK6YgkG73X799dejCbBDXz/qhC0Wi3t7e8Mblom8\nsrKyUqvVjo+P+yJdr2giyYgfdRqCHQDk1YghayOeEl8wqFarH3744dA8s7Kycu/evRE1rGzk\nlTGv+AR/MeM3IBMXCgC4kMlWqkttMMjGcifRCMFmszninOjRs8YSAgD0ScOuErOVjWAXjRBc\nXV2t1+utVqvv0VarVa/XV1dXw9ljCQEAeqU21TUajVqtFu0/0Wt7e7tarTYajRHPTWkhcdDG\nxsbW1tboc0ql0jvvvBPNIpmt1FZcAYAJTJnq4ps8MWIZkK4Ry4BkKa80Go3PPvvs888/713H\nrlgs3rx5c3l5+dq1a/GtYyfYAUBuTF+riykYTL8MiLwyFsEOAPJhJj2wMQWD6ZcBkVfGItgB\nQA7MalxdYluKXfS0bEyeAACYUmpnS3RNvwxIroJdNGdk3q0AAFIn/akuzGIZkFz1ME62QPGY\nWTBPFwoAFk3h6yE8ndlGBvGFhymXAclVsIuPMXYAkF2Fr4fO4UxfMM5gMM0yIPLKWAQ7AMiu\naYPdf9Bfnyv8MKVdebkaYwcA0Gfm5bo0e27eDQAASLfBuPTDObRiHIIdAJBPhf8yhH+Xw3Ld\niNmigh0AkBO9iafwrdD5rXk3KHHZCHYXWp0unYMZAYBYddNCoVAIjzqdX5vdS6csLo2IOtmY\nPLG/vz/vJgAA6dW3/vAsU12mZCPY3bp16/j4uFgsrq+vd84z78YCAImKfVeJ5wb+pFWWlmdr\nNptXr16tVCp37txJ+K2tYwcA6ZTEXmF/bGAdu9OUDv3KWF45ODi4ffv28fHxlStXknxfwQ4A\nUiihHWD/+ECw+75gl2WCHQCkTUKpLmQp2KW4lxgA4AzJpbqQXFyafhmQbEyeAADoSjTVJWj6\nZUAEOwAgq/KU6sIslgExdGwsxtgBQEp0y3XJfTT/ZwNj7P7PGN99mmVAVOwAgMwofCuER52Q\nu1pdrytXruzv79+9e7fZbF70uQpRY1GxA4C5K3wrzGdLif98oGL3f6Q0WarYAQAZMLdUlymW\nOwEA0m7OqS47cUnFDgBINbW68WUnggIAC6bwrRBCClJdduJSdloKACyMaE2TdE5QSDNdsQBA\nunRXqrvQFlsEFTsAIFXSuF1YduKSih0AkBZpTHWZkp0ICgDkWnpTXXbikoodADB/6U11mZKd\nCAoA5FTaU1124pKKHQAwT2lPdZki2AEA81H4VkZS3XMDf9JKsAMA5qDwrRDeyUKqyxTBDgBI\nWrT9azfMSXWzkuJiIgCQR1Gqi2Qj0mUnLqnYAQDJ6U11zFx2IigAkGWFb4UQspnqshOXstNS\nACCzFOqSIdgBAPHKfKrLTlwyxg4AiFGhUOhd1oRYZSeCAgBZ011/uFAoZGMC7FDZiUsqdgDA\n7GVmV4l8EewAgBmzq8S8ZKe2CABkQQ5TXXbikoodADAzOUx1mZKdCAoApFi0/nA+U1124lJ2\nWgoApFW0Up3ZEnMn2AEAU8l/qstOXDLGDgCYXP5TXaZkJ4ICACnT3Sus0+lE2S6fqS47cSk7\nLQUAUiOaKtG7A2w+I13WCHYAwMV0C3WkjWAHAFzAIqa67MQlkycAgHEtYqrLlOxEUABgrhY3\n1WUnLmWnpQDAnHR3lSi8Y5JEqgl2AMAofSvVFQqFhct22YlLxtgBAGey/nC2ZCeCAgDJkup+\nJDtxScUOABhCqsui7ERQACAR3V0lpLrMEewAgJ/ormki1f1EduKSrlgA4EekuqzLTgQFAOIk\n1Z0pO3EpOy0FAOLRHVQXpLqME+wAYKH1bRTW6XSibCfV/UR24lJ2WgoAzNrQ7V9FuuwS7ABg\nQQ1NdQyRnbhkViwALCKpLpeyE0EBgFnonSpBzgh2ALBAFOomkZ24pCsWABZF70p1vcuakBvZ\niaAAwBQG1x8uFAomwI4lO3FJxQ4A8s+uEgsiOxEUALg4u0rMQHbiUnZaCgBcUO9UCaluEQh2\nAJBPUt3MZCcuGWMHADkk1S2m7ERQAGAMfesPS3ULRbADgPzoW39YqpuN7MQlXbEAkBNSHdmJ\noADA2Qb3Cut0OlG2k+qmlZ24lJ2WAgBnOGsHWJFu0Qh2AJBhfVMliEV24lJ2WgoAfNlZhToW\nlmAHAJkk1SUnO3HJrFgAyB6pjqEEOwDImL5dJXqXNWHBZae2CAAL76xdJQqFggmwMcpOXMpO\nSwFgsVl/mHMJdgCQAVLdPGUnLhljBwBpJ9UxpuxEUABYSFLd/GUnLmWnpQCwYAZ3lZDqGE2w\nA4A0GlypTqqbm+zEJWPsACB1pDomI9gBQLpIdYus3W4/fvw4Wnd6Y2OjXq8PnjNiVWrrGY7F\nwo8AJGBwUN1PHioUglQ3L/+kP0UVvhnL30W73X799ddrtVrvwXK5/PDhwy+9+9m/DCp2AJAK\nUaHurB1gO52OVJd7T548qdVq+/v70V/36enp/v7+p59+urGxMeYrKESNRcUOgFgNdr+SIp8M\nVOz+aiwVu5WVlVqt1vfKrVbrjTfeuHz58ttvv33p0qUwsmInr4xFsAMgPlJd2iUV7M5KbO12\nu1wuhxA2NzdHnBYy1BXbarWisYRRmI0OHhwcPHjwIDpYrVbn20IAmIBUlwHPDfyJR6lUCiG0\n2+2+40tLS5ubm9///vfP7ZPNRiFqcCzh/v7+D37wg5WVld7T1tfXoyQ7cyp2AMzciKkSpMs/\nG6jY/ZVYKnYHBwe3b9/e3d198803Bx9ttVovvPDC+vr61tZWyHTFLhpLWKlUorGElUrl3Xff\nXVlZ2d3dPT09jUYXViqVra0tdTsAMmFwqsSINSyYs6Qqdrdu3apUKvfv3x/6m3Dp0qWjo6Mo\n1Z0lG4WowbGE0Q98enq6tLTUe7BYLO7t7c28ASp2AMyQleoy5jcGKnZ/Oca/pmaz+fHHH/ct\ncdLVarU++eST+/fvZ3jyxOAgwaHDBuNb40ewA2BWpLrsSTbYTSMbXbHFYvHcc6KRhuOcCQDz\nItVlUlJdsdPLRrC7d+9eCKE7fq5arUYBrlqtdmeOPHnypHsmAKRN4VtSHWNpNpvVanVjY6Pw\nZQ8ePNje3n769OmI52ajh9GsWAAybeiaJlJdZnw60BX7l+L6K6vVan3xZtDgJmNd2ajYLS0t\nvf/++7u7uyGEaHrErVu3isXi/v5+tOJLsVisVCoxpToAmIZUx5jq9Xq06MfJyUm0EsizZ8/W\n19fL5XKn0zk+Po6Sz9raWr1eH/oKClFjUbEDYDJSXR789kDF7i/G8he3sbGxtbXVt+hHs9m8\nevXq3t5eNA6t0Whcu3btrKKdvDIWwQ6ACUh1OZFUsDtrfY++Bd3ysKUYAGTI0KkSQarLqKRm\nxUY1ub4txaJve2cajJCrYGfNbgDSYHBXia5umJPqGBQt7lEulxuNRnSk1Wq9//77IYRopkG7\n3X78+HH320G56mGcbIHiMbNgni4UAPEZWqgj2/7VQFfsnxvreROEh2iYXd/BYrH4wQcfLC0t\nRaGlWCy+//77ly5dGnx6roJdfIyxA2AcUl0+DQt28QWDg4OD3/qt34riXalU+vmf//nu/gsr\nKyv37t179dVXe2dXfKlh8so4BDsARit8K4Qg1eVUssFuGineFAMAMkKhLueyE5ey09IQGo3G\ns2fPGo3G2tpa7/FyuXz58uWXXnppeXl5Xm0DYGFJdaRHNnoY2+32o0ePdnZ2Rp9mSzEAEjZi\nTRMfHPnx3YGu2D8zz7/fzK9j9+TJk52dnWKxeHh4eHx83Pmyk5OTo6OjYrG4tbVVrVbn3VgA\nFsXoleqswEXyslGIWllZqdVqx8fHV65cOeucaMON3nWZZ0jFDoBeZ02VsP5wPj0bqNj9qZT+\n/WYjr4xZ046v9C3YAdB11qA6qS63shPsstEVGy3f0mw2R5wTPdpd6AUA4iDVLaKkthSbXjaC\nXbTDxurqar1eb7VafY+2Wq16vb66uto9EwDiINWRgEajUavVtre3C1+2vb1drVa7u40NlZke\nxqE7bPQplUrvvPPOWWsxT0NXLABS3eL63we6Yn8mlr/r6ZcByVJeaTQan3322eeff967jl2x\nWLx58+by8vK1a9fiW8dOsANYZCN2lZDqFsLxQLD7k7H8dVer1bt37xaLxbfeeuvFF1/smzPa\narW++OKLjY2NWq1WqVTu3Lkz+AryylgEO4CFNWL9YaluUSQV7KZfBkReGYtgB7CYpDpCCOHz\ngWD3J+JahSNMtwxINiZPAEDyRu8V1v1YleqYlemXARHsAGCIcXaAjTZASqQ5zFVSy51MvwyI\nHsax6IoFWBwjpkqwoE4GumL/07gqtVMuAyKvjEWwA1gQ4xTqWDh/MBDsvhJjF/w0y4DIK2MR\n7AAWgVTHcMkGu2kYYwcAIUh15EKKdzsDgEScu/5wOmszJCc7cSk7LQWAGIyzUp0BOWSFYAfA\n4rL+MGPJTlwyxg6ABSXVkT/ZiaAAMDtSHReQnbiUnZYCwCyMXn9YqiPTBDsAFsjoNU2kOobL\nTlwyxg6ARSHVkXuCHQALQapjEWSntggAkxp/VwmpjkGdPzrvFoxNsAMgz0ZPlfjROT8u10l1\nZJ1gB0BujVmo63Q69pZghB/OuwHjM8YOgHwav/s1qNWRFyp2AOTQhVIdjKZiBwBzI9WxsFTs\nAMiPcaZKwEVlqGIn2AGQEwp1INgBkHmFT0P4n8da08QkCfLN7O6xmAYPkEKFT3/0RefmeWfa\nWIIp/N+h0HfkPymk9BdJxQ6A7Iki3bl57kcnS3UsDMEOgMwYv0T3k6dIdUzN5AkAmKULleh+\n8iypjgUj2AGQdoX2hSNdkOqYnQxV7CxQDECqFdqhs3TxZ0l1LCQVOwBSqtAOIUh1zF+GKnaC\nHQBpNFmhLkh1LDZdsQCkjlQHk1GxAyBdJk51IYROp2OHCWZOVywATGKaVBcR6Vhkgh0AqTDx\nVAmIm4odAFzA9IU6IAh2AMydVEfKZahiZ1YsAPMk1cEMqdgBMDfT7CphkgSJUbEDgHNMuVdY\n75J1QETFDoCk2SsMYiLYAZAou0qQObpiAWAIqQ5ipWIHQEKkOjJKxQ4AvkSqgwSo2AEQO6mO\nTMtQxU6wAyBG0+wAK9XBRQl2AMRlml0lpDrSI0MVO2PsAIjFlHuFdcOcVAfjU7EDYPZmsgOs\nSAcXJdgBMGMzSXWQHhnqihXsAJiZaaZKANMT7ACYDYU68ipDFTuTJwCYAakO0kDFDoBpTbn+\nsEkSpJyKHQCLYvpdJXqXrAOmoWIHwOTsFcYiyFDFTrADYBL2CoMUEuwAuDB7hUE6CXYAXIxU\nx6LJUFesyRMAXIBUB2mmYgfAuKQ6FpOKHQB5I9VB+qnYAXCOKXeAlerIugxV7AQ7AEaZfq+w\nTqdjhwlIhmAHwJlmtQOsSEemZahiZ4wdAMPNKtUBiRHsABhCqoMs0hULwJdMOVUC8idDXbGC\nHQA/oVAHmSbYAfAj069UZ5IEuZShip0xdgCEMKP1h3uXrAOSp2IHgF0lYBQVOwAyQ6qD3FCx\nA1hc9gqDcWSoYifYASyoKSfASnWQQrpiARaRVAe5pGIHsHCkOriQDHXFqtgBLBapDnJMxQ5g\nUUy/V5hUx2LKYcWuXq9vb28XCoXuv+rt7e1WqxVbwwCYpahQN+V2Yd0wJ9VBOo0V7DY2Nm7c\nuLG2ttZ7cG1t7Y033pDtANJvhjvAdjodqY5F88OBP/Fpt9uPHz+OSmkbGxv1en3wnN5CW5/z\ng93BwcHW1tb6+vrp6Wnv8f39/Vqt9sknn0zWbgCSMcNUB8Sq3W6//vrr9+/fj77d2tq6cePG\n9vb2+K9wfrB79913Qwhra2tLS1+6Mdy6dSuE0H1vAFJIqoPpJVaxe/LkSa1W29/fj0rjp6en\n+/v7n3766cbGxpivUDi3oh7V+qLTer8e/DbHCoXzLxRA2kh1MBObob/f878rxJJ/VlZWarVa\n3yu3Wq033njj8uXLb7/99qVLl8LIAHZ+xa5UKoUQ2u123/Fms9l9FIBUKbSlOsieWq02ePDS\npUsffPDBV77ylV//9V8/9xXOD3avvfZaCKFcLvdmu3q9vrq62n00Ae12u1qtrqysdI/UarUH\nDx50Rxc+ffo0mZYApNyUE2BHjMuGxZRYV+xZ1bSlpaXNzc3vf//75/bJjtXDuL293TclNrK+\nvr65uTl2aycXFSGjGNvpdKKhhYOpNr726IoFssL6wzBzf2egK/bvxtMVe3BwcPv27d3d3Tff\nfHPw0Var9cILL6yvr29tbYWJu2JDCA8fPjw6OiqXy90j5XL58PAwmVQXQvjkk09qtVq5XI5m\n5n700Ue1Wm13d/fk5CQaXXhycrK7u7u1tVWtVpNpEkAKSXUQh8Qqdrdu3apUKvfv3x9aNb90\n6dLR0VGU6s6SjUJU9OOdnp5GM3P7vo202+3nn3++WCzu7e3F0YBMXChgkUl1EJO/PVCx+3vx\nVOwizWbz448/fvjw4dBHW63WJ598cv/+/aENyEZe6Zv9MXTOyOBps21AJi4UsLCkOojPfzsQ\n7P77OIPdNM7vih0xinZlZaV3NkN8dnd3w4/n4YYQoo7n7reRaKTh+vp6Au0BSI/pJ8BKdZAb\n4+4VO6jVatVqtaHzcmfuG9/4RghhdXW1Wq02m81isVgul1dXV7vZrl6vP3r0KITwzW9+M4H2\nAKTE9DvASnVwriS3FGu1WtGWYlH/ZHTw4OAgWglkZWVl9HSC4T2M9Xr9xo0b47x9qVR67733\nJmj3RdXr9Y2NjdE5cm9vr1gsxvHuumKBFJp+pTqpDsbxcKAr9u/H0xU7uO7H/v7+D37wg74O\n0hHLgAyv2F2/fn2clYdLpdKv/uqvXqTBk7t+/fre3t7h4eHu7m5f28rl8t7e3snJSUypDiCF\npDrIn2hLsUqlEi36UalU3n333ZWVld3d3dPT02iTsUqlMmIZkIttKbawVOyAVJHqIEl/a6Bi\n9w+S2lJs6EoghULhrGVA5JWxCHZAShTaIYTZ7BXm/+0wpsSC3eC/yqH/TqfaK3aERqOxvb09\nzSsAML7pp0r0ivp6ZvNakGuJTZ4YZ1BZtAzIWWeOFeyq1WphmGvXrg3damxe7G8I5Nj03a9A\nyt27dy+E0B0/V61WowBXrVa7G8g+efKke+ag83sYq9Xq3bt3z3r06Ojo+vXrF295LCbrVhgz\nC/p/LTBHUh3M0d8c6IrdHa+OdNHwENes2F4ffvhhCOHw8LDT6UTL/56cnJycnERf/9RP/dSF\nWhyryboVOmOIo7UAY5LqIIXiyA9LS0vvv/9+tC9DND3i1q1bxWJxf38/WhKkWCxWKpWzUl24\n6KzYWq22srISVemazebVq1dHZMY8MXkCmBepDububwxU7P5hdrcU63XlypUQwh/8wR90v97a\n2oqjWQDMZK8wI49hoZwf7Mrlcgjh6dOnIYSvfvWrIYSPP/44hNBoNGJuW79Go1Gr1ba3t/vm\ncGxvb1er1eTbAxCfGe4VJtvB4ji/h7G7vVh05sbGRm+Vbnd3980334y1iSGEdrv96NGjnZ2d\n0afF1y+sKxZIkvWHIVV+eaAr9h/NtSt2qnXsor28ut9ubm5GY/pCCJVKJYFUF0J48uTJzs5O\nsVg8PDw8Pj7uG5l4cnJydHRULBZH7LABkBVSHTCxbBSioh02jo+Po4F9Q0WTOc7aYWNKKnZA\nMqQ6SKG/PlCx+zCtkyfOzyvb29uXL1++c+dOMg0aaswF6uLbHkewAxIg1UE6ZSjYnd8Vu7a2\nNmKB4mREyy43m80R50SPjrMXB0DaTD8BNkh1EJvEthSb3vnBLloQr7uRxVxE+2asrq7W6/VW\nq9X3aKvVqtfrq6ur4ewdNgBSayY7wEp1kBvTLANyfg9ju91+//33T09PX3/99eXl5Zm2/AL6\nZuMOVSqV3nnnnaWl2S/lqSsWiMlM1h+W6iBWf22gK/afxtMVO/0yIOPuPDFCYjeRRqPx2Wef\nff7552tra92DxWLx5s2by8vL165diy93CnZAHKQ6yITEgl21Wr17926xWHzrrbdefPHFvjmj\nrVbriy++2NjYqNVqlUpl6PyHLAW7ORLsgJmb4V5h8U0dA0KCwW76ZUDklbEIdsAMFdohBDvA\nQmb84kCw24sn2E2/DMjF9ooFYEozmSoB5NL0y4AIdgDJmWH3K5CYxJY7mX4ZkOdiaxsAXyLV\nAaPduXPnu9/97tbWVq1WG3FaqVR69dVXhz5k6NhYjLEDpiTVQXZ9Y2CM3T+Pc+eJaZYBkVfG\nItgB05jVrhJuRDAXCQe7aeiKBYjRTCbAdped8p9MmIs07yHWR7ADiIv1h4GEjTsrtl6vd/cs\ni45sb28PztcAICLVAckbq6rft09r9JRCoVAsFt9///1Lly7F2MB00P0BXIhUB3lyc2CM3f+S\n1jF251fsDg4Otra21tfXT09Pe4/v7+/XarVPPvkktrYBZJJUB8zL+YWoaNuy09PTpaWlvmlZ\nizNLS8UOGMes9gqT6iBVXhmo2P1OWit25+eV3vQm2AGcZVYr1Ul1kDYZCnbnd8WWSqUQQrvd\n7jsebVUWPQqw4KQ6yLHEthSb3vnB7rXXXgshlMvl3mzX3aosehRgkUl1QEqM1cO4vb3du6lF\n1/r6+ubmZgytSh1dscBZZrtX2OIMcYEM+S8GumI/S2tX7Lh5pV6v/+Zv/mY33pXL5VdeeeXl\nl1+Os20pItgBQ9kBFhZBDoPdghPsgD6zmgALpF+Ggt35Y+wGp00ALLioUCfVwYLI1eSJ559/\nfmNjo16vJ9AagPTT/Qqk1ljLnWxtbd24caNQKFSr1UajkUCzANJJqoMFlKGK3VhDx1qt1r/4\nF//iO9/5zs7OTgihVCr9/M///M/93M8tLS3K7c0YOyDMdK8wtxTIkD89MMbuX6d1jN3F8kqr\n1To4OPjwww9rtVoIYX19/Zvf/Ob169dja15aCHaAHWBhYV0bCHaNfAS7roODg9u3b0dfp/MH\nmy3BDhaZHWBhwWUo2J0/xq5Xq9Wq1WoPHjyIUl2pVDo8PIynYQCpMKsJsFIdZFeGxtg9N85J\nfWPsisVipVK5devWpUuXYm4ewDzZKwzIlvOD3YMHD6I8F0LY3d29efPm8vJyzK0CmD+pDsic\n84eOFQqF9fX1X/iFX1icDcQGGWMHi0aqA7quDoyxa6Z1jN35FbvT09PFWdYEYIZ7hUl1QMKG\nB7veZZaef/750S/hbgXkxgzXH5bqIDfSPFuiz8VmxQLkWEy7Skh1QGKGV+x6b0NuScAimG2q\n65br3EIhB1TsALJk5rW6KM9JdUDCzp88MWJbw5WVlRDC3t7ezJsFkBg9sMBoGarYjbXcSRh2\nh2q1Wi+88MLQh/LHcieQSzOcAAvk2E8PLHfyb9K63Mnwrth6vV74sehIYUCU6kqlUnKNBZid\nWe0VBpAew4Pd9evXx0lspVLpV3/1V2fdJIDYxdT9CuRShvaKnbwrdqHoioU8kRONXlYAACAA\nSURBVOqAC1ka6Ir9t2ntij1/8kQ62w0wmZkvQewmCbmX5hJdn/N3nuhdPH0oNzUgK+LYWEJF\nH0iP8yt2ADkw2wmwtguDhZL5ip2dJ4A8iWlXieAOCaSMih2Qc1IdMKUMVezG2lLs4ODgwYMH\n0detVuvBgweFQuHBgweNRiPOtgFMS6oDFsr5Y37r9fqNGzfCj+9iKysrtVqt++izZ8+Wl5dj\nbWIaGBwNWSTVATPxRweWO/n3aV3u5PyK3T/+x/84hHB4eBhCaDabtVqtVCp1Op3oyAcffBB3\nEwEuqtCW6oCZ+ffhj/T9mXeLznSxBYqbzebVq1crlcqdO3fCIq3hpGIHGTLz9YelOlhwhfBH\nBw79+3TeDS42eeLjjz8OIfzsz/5sPI0BmJZUB8RgMC/9v3NoxRjOryWur6+HEJrNZr1eX1tb\nKxaL169fDyE8ffq0+yjAXBTCb33pTwx7hXXDnFQHpN8FJk9E9vb2isVi+PH/Yo+OjqKcl2+6\nYiGFCuG9Tngw71YA+VcI/9HAoT9MZzA4v2J3/fr1/f39KMxVKpXoixBCqVQ6PDxchFQHpE0h\nvCfVAQxSiBqLih2kh0gHJKwQBgZ5FP5tOoNBeufrAgyS6gBGGDfYPX36dGNjo/Bj29vb0eQJ\ngMRIdQCjjdXDuLGxsbW1NXh8fX19c3Mzhlaljq5YmLtYU93irMoJTKAQfnrg0L9J5x3j/Ird\nwcHB1tZWsVg8Ojrq/NjR0VGxWNza2jo4OEiglcAii3uqRHelut4l6wCy6PxgFy1K/O1vf7t3\nAuz169e//e1vdx8FiEkU6RJIdUHFDjjTcwN/UupiW4qN/1DO6IqFuYh7UJ1UB4yjEF4YONRK\n503j/IpdqVQKITSbzb7j0ZHusnYAsyXVAamRmYrd+cHutddeCyGsrq7W6/XuwXq9vrq6GkL4\nlV/5lfgaByymBNYfluqAXDIrdiy6YiExCaxpItUBF1IIVwcONdN59xg3rzx9+vS3f/u319bW\nom/L5fIrr7zy8ssvx9m2FBHsIBlSHZBCOQx2C06wgwRIdUA6ZSjYpXf0H7BQpDogxTKTly6w\npdj29nZ3S7HHjx/3zqUAmFgCUyUi3TAn1QF5NdXkiXK5/PDhwxhalTq6YiEmtn8F0q8Qrg0c\naqQzGJxfsatWq2dtKba2tmZLMWBiUh3AbJ0f7D788MNgSzFg1qQ6IDsys0CxLcXGoisWZqgQ\n3gshSHVAVhTCnx449K/TGQzOr9hVKpVw9pZi5XI5jmYBeRUV6qQ6IFMyU7E7P9jduXNnb2+v\nb0uxRqOxurq6vr6+IJMngJlIZk2T3mVNABbKuF2xY0pnWXJ6umJhelaqAzKqEK4PHPrf0nmT\nSW8tEcgTqQ7IsszkpfMb6v4ITCOZqRJSHUDIUAIFsiiZNU2kOiBmmclL424pBnBRUh1AwjKT\nQIFskeqAHMlMXspMQ4GsSGz9YakOoI9gB8xSYhuFSXVAgjKTl4yxA2ZGqgOYL8EOmI3EUl3o\nCXNSHUCvcYNdvV7f3t7u3atne3u71WrF1jAgS5JMdZFOpyPVAUnJzF6xY+2UtbGxsbW11f02\nekqhUCgWi++///6lS5dibGA62FIMzpLYVAmAeSmE2wOHDtIZDM6v2B0cHGxtba2vr5+envYe\n39/fr9Vqn3zySWxtA9IuKtRJdUDe5ahit7KyUqvVTk9Pl5aWon7Y7lP6vs0xFTsYlHz3K8Bc\nFMI3Bg7983QGg/PzSm96E+yAiFQHLI5C+IWBQ99JZzA4vyu2VCqFENrtdt/xZrPZfRRYHIXw\nXmK7SvQuawKQe4WLGPoK5we71157LYRQLpd7s129Xl9dXe0+CiyIxAbVde9Zsh2QAgmNsdvf\n35/yFcbqYdze3l5bWxs8vr6+vrm5OWULppdAj7CuWAjWHwYWVSH84sChvZjuTs1mc3V19fr1\n65NFrHHzSr1e/83f/M1uvCuXy6+88srLL788wVvOnGAHCZDqgIWVZLALITSbzatXr1YqlTt3\n7lz0udnIKxfqi4njJxLsWHBSHbDICuGvDRz6p7Heow4ODm7fvn18fHzlypULPTEbW4o9e/as\nWCzOuxWwiBKbKhGkOoAfu3XrVqfTuWiqC1kJdsvLyx988EG5XA4hrK+vn5ycdHpE5/R9C0wv\nyfWHpTogxXK0QPG53aBJ3oKfPn369a9/PYRweHjYHeFnjB3EIcmV6qQ6IM0K4b8eOPRROm9W\n2ajYdb388ssnJyfr6+tf//rXt7e3B1fXA2ZCqgPokVDFrlAorKysNBqNyV9hsttoo9H44IMP\nnn/++YcPH0783tOo1WorKyvFYnFzc/PGjRtBxQ5mpBDeCyFIdQBdhfDXBw59GNNkzeiLcrn8\nxhtvLC0tXfQVJqzYLS8vr62tra2tPX78eLJXmFKxWDw+Pr58+XKU6oCZSHJQXaR3mGxibwpw\nQYmOsTs6Ovr000+ff/75arV60c7JqQpRadgr9vHjx/fv34+7GSp2LALbvwIMVQi/PHDoH8VX\nset0Ou12+/3334/WD97d3f3a1752/fr1sV5h4mZFK6wUi8W9vb3JXiFDBDtyT6oDOEvywS76\nttlsfvzxx73bQ/zZP/tnf+ZnfmZpaemsXtppZ8Xu7e0twgpzgh35JtUBjFAIf2Pg0D9MINhF\nms3m7/zO73z44Ye1Wq33+NAGTB7sSqXSa6+9duvWrYs1OU7xdQ0LduRVwlMlALJovsGuq9ls\nfu973/vd3/3d733vezs7OxMGuwyZLNiNuV9Zni4URBTqAMZRCH9z4NDuOE+cLJNMEznOnxW7\nvb1drVYnfoMkTbbzRGcMcbQW5ivhleoutOMzQPqlMz+M2xW74OFGVyw5Y/1hgPEVwn8zcOh/\nSOcN7fyKXalUCiHY4wFyQ6oDyKvzC1HRSiqnp6evv/768vJyMs0aqtFoPHv2rNFodOf9Rsrl\n8uXLl1966aX4mqdiRz7YVQJgAoXwtwYO/YN03tamXe4kJHK/brfbjx492tnZGX3a+vr65uZm\nHA0Q7MiBhKdKSHVAbqQt2I0YJjfhlmIJe/Lkyc7OTrFYPDw8PD4+7huZeHJycnR0VCwWt7a2\nsjLPAxIm1QEsgmwUolZWVmq12vHx8ZUrV846p9lsXr16NaadMFTsyDSpDmAahfBw4NDfT+f9\nbXheSdtM2DHbY4Fi6JP8+sNSHZA/GQp22eiKjXYtazabI86JHl2E/c1gTFGhTqoDmNpzA39S\nKhvB7t69eyGE1dXVer3earX6Hm21WvV6fXV1tXsmkPyuElIdwEw0Go1arba9vV34smjPiEaj\nMeK52eiKDSFsbGxsbW2NPqdUKr3zzjtLS0szf3ddsWTLvPYKS+GtA2B6hfC3Bw79vTjuddMv\nA5KZYBdCaDQan3322eeff967jl2xWLx58+by8vK1a9esYwfBDrAAs5ZYsKtWq3fv3i0Wi2+9\n9daLL77YN2e01Wp98cUXGxsbtVqtUqncuXNnSFNHBLsxLULiEezIhOSnSgAsgkL4OwOH/m4c\nwWD6ZUAEu7EIdqSfQh1ATBILdtMvAzJq8kRnPFO0H5gNqQ4gB6ZfBiQbs2KBEaQ6gJgltNzJ\n9MuApHchFuBc81p/WKkeIA537tz57ne/u7W1VavVRpxWKpVeffXVoQ9laVbsHBljRwpZqQ4g\nGYXwawOH3orvNjjNMiCC3VgEO9JGqgNITMLBbhrDu2LT2VYgItUBJCszQ9dMnoCMkeoAOEtm\nEigwl/WHpTqADOWlzDQUFtxc1jSR6gCyRVcsZIBUB8A4VOwg7aQ6gHnLTF7KTENhAc1lUF2Q\n6gAyS7CDlJrjRmGdTsdilgA9MpOXMtNQWChz3/5VpAPIIsEOUmfuqQ6AL8tMXjIrFtJFqgNg\nYplJoJB785oqAcB5MpOXMtNQyDeFOgCmpysW5m9eK9X1LmsCQA6o2MGczXf94UKhYAIswHky\nk5cy01DIH+sPAzBbgh3Mx7wG1Ul1ABeXmbxkjB3MgVQHQBwyk0AhN6Q6gKzJTF5SsYNESXUA\nxCczCRSybo7rD0t1ANPJTF7KTEMh0+a4/rBUB7A4dMVC7KQ6AJKhYgfxmu9eYZ1OJ8p2Uh3A\nFDKTlzLTUMicOQ6q6yXSASwOwQ5iMd9CHQAzlZm8ZIwdzJ5UB8BcZCaBQlZIdQC5k5m8pGIH\nsyTVATBHmUmgkHJzX3/YJAmA2GQmL2WmoZBmaViprlAoyHYAC05XLEwrDakuqNgBoGIHs/C9\nubyrVAeQlMzkJRU7mEohPOyE7Tm8r1QHwIDMJFCgS6oDSFZm8pKKHUxuLuU6qQ6As2QmgQJB\nqgOYj8zkJRU7mFDy5TqpDoDRMpNAgS6pDiBZmclLKnaQDd1ynVQHwFkEO5hE8v2wUZ6T6gAY\nITOlRUCqA5iTzOQlFTu4sHktSgwAo2UmgQIAzElm8pKKHUzgf513AwBgiMwkUACAOclMXspM\nQyElCuFqJxzH9eKFQjBJAoBJCXZwAQmkuugL2Q4gTTKTl4yxg3Elk+qCih0AkxLsYCxSHQDp\nl5nSIsyRVAew2DKTl1Ts4BxSHQBZkZkECnMh1QGQobykYgdnkuoAyJbMJFBImFQHwI9lJi+p\n2MEQUh0AWZSZBAq50el07DABkCmZyUuZaSgkJtZyXUSkAyAOumLhSxJIdQAQExU7+AmpDoBh\nMpOXVOzgR6Q6ALIuMwkUYiXVAXC2zOQlFTuIJdUVCoXeZU0AIAGZSaAQk5hSXfcLE2ABsi8z\neUnFjoUWa6oLljUBIFmZSaAwc1IdAOPJTF5SsWNBSXUA5I9gxyKS6gDIpcyUFmFWpDoALub/\ny8wqByp2LBapDoAcU7FjgUh1AEzih/NuwNhU7FgUUh0Auadix0KIacewTqcTZTupDiDPslOx\nE+zIv1j3gRXpAEgPwY6cizXVAbAQslOxM8aOPJPqAFgogh25JdUBsGh0xZJPUh0AM6MrFnKj\nUCj0LmsCAKmlYkcOzbBc1410hULBBFiABZVUxe5CdYShn0oqduRNHKkuWNYEgPjt7+9P+QqZ\nCXbtdvvx48dRp9jGxka9Xh88R5cZUh0As/fDgT/xuHXr1vHxcbFYXF9f75xn6Ctko3ep3W6/\n/vrrtVqt92C5XH748GHvkfj2ANANlwlSHQBxKHw+cOhPxBgMms3m1atXK5XKnTt3LvrcbFTs\nnjx5UqvV9vf3o4h6enq6v7//6aefbmxszLtppIVUB0BckqrYRa5cubK/v3/37t1ms3nR52aj\nELWyslKr1fqa2mq13njjjcuXL7/99tuXLl0KKnYLTKoDID6FwU+YP5nSYJDSZvU5K7G12+1y\nuRxC2NzcHHHaTBqQiQu1mKQ6AGKVoWCXja7YUqkUQmi3233Hl5aWNjc3v//97+uTXVhSHQCx\nS7YrdhrZCHavvfZaCOGjjz4a+ujbb7+9tbUl2y0gqQ6AnGm329VqdWVlpXukVqs9ePCguzDI\n06dPRzw9pYXEQdVq9e7du+GMD916vX7jxo3oa12xC2Lmm4bF15UPQKYVng0c+lOxBINo/kC0\nDEin0xm6KkgIYX19PRqENigbFbsQwp07d46Pj6MRdYOuX79+cnKyu7ubcKuYlzi2gh2xLBAA\nJOCTTz6p1Wrlcvn09DSE8NFHH9Vqtd3d3ZOTk+hDKko7W1tb1Wp16CsoRI1FxS5V4kh1AHCW\nwncHDv2ZWIJB1Hd0enq6tLQ0+G2k3W4///zzxWJxb29vyCvIK+MQ7NJDqgMgYQkHu+4rD13u\nbfC0XpnpioUg1QEwF0nNio0GlXXXJX7zzTd7v41Ei4Ssr68PfYVcBTt7xeabVAdAvn3jG98I\nIayurlar1WazWSwWy+Xy6upqN9vV6/VHjx6FEL75zW8OfYVc9TBONqtxzCyYpwuVRTNJdea9\nAjCBwr8aOPTn4goP9Xp9Y2NjcCZsr729vWKxOPShXAW7+BhjN18zTHURf5sAjG9osIv1o+Tp\n06e/93u/99lnn+3s7HQPlsvl5eXlP//n/3y0k+pQ8spYBLs5kuoAmK/kg93Enpt3A2AUqQ6A\n+UvxHmJ9shTsGo3Gs2fPGo3G2tpa7/FyuXz58uWXXnppeXl5Xm0jDlIdAFxISguJfdrt9qNH\nj3q7mYcascPGlHTFJk+qAyAlCr89cOgvzjMYZH4duydPnuzs7BSLxcPDw+Pj486XnZycHB0d\nFYvFETtssICkOgAWTTYKUdHKy8fHx1euXDnrnGazefXq1bN22JiSil3Cpi/XSXUAzErh04FD\nfymlwSClzeoz5vJj8a1SJtglSaoDIFUyFOyy0RUbrcLXt6VGn+jRs9brIyukOgBSJ6ktxaaX\njWB37969EMLq6mq9Xm+1Wn2Ptlqter2+urraPZOMmsmEiW6Yk+oAyKJGo1Gr1ba3twtftr29\nXa1WG43GiOemtJA4aGNjY2tra/Q5pVLpnXfeWVpamvm764pNgK1gAUinwv7Aof8qlmAw/TIg\nWcorjUbjs88++/zzz3vXsSsWizdv3lxeXr527Vp869gJdnGT6gBIrcJvDBz6y7EEg2q1evfu\n3WKx+NZbb7344ot9c0ZbrdYXX3wR7SRbqVTu3LkzpKnyyjgEu1hJdQCkWWLBbvplQOSVsQh2\n8ZHqAEi5wj8bOPRXYgkG0y8Dko3JE+SVVAcAXdMvAyLYMTfTpLpoftBs2wMAwyW13Mn0y4A8\nF1fTYKQpU133C13kAOTGnTt3vvvd725tbdVqtRGnlUqlV199dehDPhfHIkDM1kxSXbBSHQCJ\nKPyTgUPfjDEYTLMMiLwyFsFuhqQ6ALIl4WA3DWPsSJRUBwDxMcaO5Eh1AGRSijeH7aNiR0Kk\nOgCIm4odSZDqAMgwFTvokuoAIBkqdsRLqgMg81TsIEy9Y1g3zEl1ADAOFTviMpN9YEU6AOZP\nxQ4AgIQJdsRiJuU6AOBCdMUye1IdALmiK5aFJdUBwLyo2DFLE6S6aE0TkyQASC8VOxbQxKku\nfHnJOgBgMip2zMY0qS6o2AGQZip2LBSpDgDSQMWOaUl1AOScih0LQqoDgPRQsWMS3wiFEMI/\nD1ekOgBID8GOC/vlUPiNMEkmk+oAyCRdsdBHqgOAuKnYcTG/HAr/48XLdVIdABmmYkcuTZbq\nQk+Yk+oAID4qdoxr4lQXEekAyCoVOwAAEqZix1imLNcBQIap2JEnUh0AZIJgxzmkOgDICl2x\nzEy0polJEgDkja5Y8mH8cl13pbreJesAgCSp2HGmCVJdULEDIH9U7Mg6qQ4AMkfFjqlIdQDk\nn4odmTZmuU6qA4BUUbGjn1QHAF+iYkdGSXUAkF2CHRcm1QFAOumK5SfGKddJdQAsHF2xZM6Y\nnbDdMCfVAUDaqNgRwgU3hBXpAFgsKnYAACRMxY6LlesAYOGo2JEVUh0A5IaK3UKT6gDgfCp2\nZFqhUOhd1gQAyATBbnGdVa7rRjrZDgCyRVfsgjo31QXLmgBARFcsaSbVAUAuqdjxI1IdAAyn\nYkdqDS3XSXUAkAMqdotFqgOAC1OxIyukOgDIDRW7BTJYrpPqAOB8KnakzehNJqQ6AMgBwW4h\njB5aJ9UBQD7oil1cnU6nUChIdQBwDl2xpMeITlipDgDyRMUu50YPrQMAzqdiRxpIdQCwUFTs\nAABGUrFj7pTrAGDRqNjlUzfVRWuamCQBAJNTsWOO+lJd+PIOEwBAXgl2uWW7MABYNLpi8yYq\n10l1ADAzumKZC6kOABaZil1+SHUAEAsVO+ZCqgOARaZilxO/HAr/qGfmq1QHADOjYkeSpDoA\nIKjY5YBUBwDxUrEjSd0wJ9UBwCJTscu27iYTIh0AINhlWDfVAQAx0hVL3KQ6AKCPih0AwEgq\ndsRKuQ4AGKRilxnRrhKdTudQqgOAJKnYMVvdvcJ6Nw0DAOilYpcBvWHuX3fCzyrXAUCSVOyY\nFakOABiTYJdqUh0AML4sBbtGo1Gr1ba3twtftr29Xa1WG43GvBs4Y1IdAKTCDwf+JOjg4ODB\ngwfdzFOv10ecXMjEVlTtdvvRo0c7OzujT1tfX9/c3IyjAYVC0hdKqgOAlCh8deDQ/xVLMOiu\ngNE9srGxsbW11XdapVK5c+fO0FfIRsXuyZMnOzs7xWLx8PDw+Pi482UnJydHR0fFYnFra6ta\nrc67sTMg1QFAisypYletVre2tiqVyunpaTfzVCqVu3fvHhwcDH1KNip2KysrtVrt+Pj4ypUr\nZ53TbDavXr1aLBb39vZm3oAkK3ZSHQCkSuGPDxz6fhIVuyj/nJ6eLi0tdc9pt9vPP/98qVR6\n7733hrxCJoLdYGVymtMma0AyF0qqA4C0KfyxgUOnSQS7s4LNiMCTja7YYrEYQmg2myPOiR6N\nzsyu7l9SJgI3ABCf3d3dsx46K/BkI9jdu3cvhLC6ulqv11utVt+jrVarXq+vrq52z8y0qBP9\n90NBuQ4AUiHZMXbb29u1Wu3g4OBrX/taCOHp06fdh9rtdjSd4KzAk42u2HDGrJA+pVLpnXfe\n6e2HnpWEZ8VKdQCQHoX/eODQ/xNjV+ygvs7ZYrH4wQcfDA08mQl2IYRGo/HZZ599/vnna2tr\n3YPFYvHmzZvLy8vXrl1bXl6O6a2TDHZSHQCkSmLBLvx4aNn3vve9EMLv/u7vhhDW1tZ6p1Pc\nu3fv1VdfPauMlaVgN0eJBTupDgDSpvBTA4f+XUoTVDbG2C0IqQ4AmMZz827AIho6S1mqA4CU\nSnYPsWnkqmIXbaM271aco9vC3qZKdQDAmEYEnpT2EE9msgWKx8yCM7lQve/V+Z9CCCH8h+H3\nv2EhYgBIr2FJIbnwMOS9zw48uQp28ZnJ5IkvpTqXHQAyYmiwS+dHea66YtNMqgMA4mbyRBKk\nOgDIrsGP7tQO6c9SsGs0Gs+ePWs0Gr0LFIcQyuXy5cuXX3rppfgWKJ6GVAcAjG+awJPSHuI+\n7Xb70aNHOzs7o09bX1/f3NyMowETj7GT6gAgf2LauWD6wJONMXZPnjzZ2dkpFouHh4fHx8ed\nLzs5OTk6OioWi1tbW9HOuCkh1QEA45s+8GSjYreyslKr1Y6Pj69cuXLWOc1m8+rVq8VicW9v\nb+YNmCCYS3UAkFcxVeymDzzZCHZjLlA32Tp2YzZggpeNrz0AwBzFFOymDzzZ6IotFoshhGaz\nOeKc6NHozJSICqfzbgUAkA3TB55sBLt79+6FEFZXV+v1eqvV6nu01WrV6/XV1dXumQAAmTN9\n4MlGV2wIYWNjY2tra/Q5pVLpnXfeWVpamvm7x1RxBQCyKL5gMGXgyVJeaTQan3322eeff967\nrEuxWLx58+by8vK1a9fiW8dOsAMAumINBtMEHnllLIIdANCV2mCQjTF2qVUoFAqp3VUEAFgw\ngt3kupFOtgMA0kCwm5D1hwGAtBHsJiHVAQApJNhdmFQHAKSTYHcxUh0AkFqC3QVIdQBAmgl2\n45LqAICUE+wuTKoDANLpuXk3IHusWgcApFNKN8TIsdRuQrIgXP/5cv3ny/WfL9d/vhbk+uuK\nBQDICcEOACAnBDsAgJwQ7AAAckKwAwDICcEOACAnBDsAgJwQ7AAAckKwAwDICcEOACAnBDsA\ngJxYiH3TAAAWgYodAEBOCHYAADkh2AEA5IRgBwCQE4IdAEBOCHYAADkh2AEA5IRgBwCQE4Id\nAEBOCHYAADkh2AEA5IRgBwCQE4IdAEBOCHYAADkh2AEA5IRgN2PVanVlZaVQKKysrFSr1Xa7\nHcdTOMsEF/Pg4ODBgweu/0xM88vcaDQKhUJ8bVsEE1z/ZrP5+PHjQqFQKBQeP37carUSaGde\nXfT6t9tt9/+ZOzg4GP9Oks/r32F21tfX+y5vqVSa+VM4ywQXs1Kp9D2lWCyenp4m0+CcmeaX\n+fT0tFgsuiNNY4Lrf3R05Pd/Vi56/U9OTqLf+b7rf3Jyklib8+f4+Hj8bJPXz1+30Zl59uxZ\n9M/y+Pi40+kcHx9H/2ifPXs2w6dwlgkuZnQL2N3d7X6S7e/vR0cSanSOTPnLXC6Xx78dM2iy\n618sFrtPOT09jf4WKpVKQo3OkQmu/+7ubt/Vjv6f6f4zse5lH+dOkuPPX7fRmYn+TR4dHXWP\nHB4ehhD29vZm+BTOMsHFjD7G+g6KF5OZ5pc5OtOVn8YE1z86oTdYnJ6eRh918bY1jya4/kN/\n4f0rmFj03/JuEe7c83P8+WuM3cx8/vnnIYSvfvWr3SMvvvhiCKHRaMzwKZxlgov58OHDzo/v\npExp4l/mVqv19a9/fbBPhAuZ4Pr/3u/9XgjhL/yFv9A9srS01Ol09vb2YmxoTk1w/Qf7YUcf\nZ7Tbt29XKpXNzc0xz8/z5++8k2V+DL2eoy/yBE/hLDO5mFHFYn19faZNWwgTX/8o0p2cnPjN\nn8YE179UKkWP7u3tRWGid1gCFzLB9Y8C9GBXbA4qRnMR9ah2xr7t5/jzt9BRsZiRaBpO3/Uc\nenCap3CWmVzMg4OD27dvHx0dXb9+feYtzLfJrn+tVltZWTk8PHz55Zf95k9j4vtPuVxeW1vr\nHiwWiyp2E5js9//g4ODdd9+t1WrRt8Vi8Vd+5Vdu3boVZ0vzb8w7SY4/f3XFwo+0Wq133323\nXC5LdcloNpsrKyvlcvnll1+ed1sW16efftqdPLG7u1ur1bo5g7j94Ac/6L3atVrtBz/4wRzb\nQz4IdhBCCO12+4033rh+/frDhw/n3ZZF8Wu/9mvFYvGNN96Yd0MW2re//e0rV66EEJaWln7x\nF38xhPCd73xn3o1aCAcHBysrK92u2ChYr6ysHBwczLtpZJtgNzMTjHg1SHaGpryY5XL58uXL\n4w+8pc9Fr//jx493dnY2NzeXlpZiatJCmfj3P0p1kUuXLoUQdnZ2ZtOmRTLB9f/4449DCK++\n+mr07dLS0i/90i91jxO3HH/+CnYzc/PmzRBC77rt0dfdBbpm8hTOMvHF0jOCCQAABvpJREFU\nbLVaDx48CCG8/fbbcTYw5y56/e/fvx9CuHHjRuHHouO9XzO+CX7/3WdmaILrHwXo3v/YRF8L\n1snI8eevYDczy8vLIYQvvviieyT6+vLlyzN8CmeZ7GLW6/UXXnjhK1/5yubmZlSuYDJ+medr\n4vtPvV7vHok+2Cw9M4EJrv9gxSjazyrHlaRUyfMtK8kpuPlm54n5muBiRktsWNxkJqb/ZXZH\nmsbEv/+lUila4iQa4xVCODw8TK7deTHB9Y8WN6lUKt0lZrpHEmp0To15J8nx56/b6CxF60L1\n6gsNg79w5z6F8V30+kcfY/7DMysT/P6P/yjnmuD6D65s4v4zsQmu/2Bx1PWf3ll3ksHjef38\ndRudpdPT00qlEqX+YrE4+B+vwV+sc5/C+C56/Qfi3E8k2/CcmOD3f/xHOddk1//o6CiKF+4/\nU5rs+rv/z9z4wS6vn78WKAYAyAmTJwAAckKwAwDICcEOACAnBDsAgJwQ7AAAckKwAwDICcEO\nACAnBDsAgJwQ7AAAckKwAwDICcEOACAnBDsAgJwQ7AAAckKwAwDICcEOACAnBDsAgJwQ7AAA\nckKwAwDICcEOACAnBDsAgJwQ7AAAckKwAwDICcEOACAnBDsAgJwQ7AAAckKwAwDICcEOACAn\nBDsAgJwQ7AAAckKwAwDICcEOACAnBDsAgJwQ7ICpFM52oVeItZHx6Wt8vV4f8Wga9LUQyBnB\nDmA2Hjx4cOPGjXm3YpT0txCYUqHT6cy7DUCGRRWpae4k079CSqT/B0l/C4EpqdgBAOSEYAfE\nrl6vb29vRwPOVlZWqtXq6JM3Njaikzc2NgbHhB0cHDx48CB6qYODg9Fv3R3lVq1WR7x79zUf\nPHgw+JojmtR9/e5Yut5xddHX7XY7euW+l43esd1uT/xztdvt6Cnb29vdpp51qYe2cIK3BlKt\nAzCFc+8ke3t7g3eeSqUy9BX29/cHT97f3++eXC6X+x5dX18/t3l9beh7yvr6+ojXHN2kbuOH\n3lq7X0dvcXJy0n3Zk5OT3jea7OfqPiu6nqMv9dAWTvDWQJoJdsBUBpNEX26Ivj08PIy+PTw8\nHHpC9HWxWAwhHB8f955cKpWib4+OjqLYcXp62ul0Tk9Po8B0dHQ0unnFYjF6zePj4+gt+toz\n4oTRTRoa4wZ/tOhZe3t73YeiEBa1fOKfq/uUCS71xJcUSDPBDpjKucHurKcM/Tb6+qxUEdWW\neqPM6elpCKFcLo9+r94XjKJM9ynRaw6e0K1ajW7SmMEu+robBzudTqlU6j408c/17Nmzs04Y\n2qTBFk7w1kCamRULTGXMiZatVuuLL75oNpv/8l/+y62trd6n9L5CtVq9e/duCKFcLr/22msh\nhCtXrvS911BnNWBo83oPnnvCOE0666V6jzx+/Pj+/fsnJyeXLl1qtVovvPBCpVK5c+fODH+u\nyJiXuvfIhd4aSLXEoySQK+PcSQYHsYWzy0h7e3tR72ekWCx2h6ZNcCsb+mjvwXNPGKdJZ71U\n75Fnz56FH/fGViqV0FNvm9XP1bngpZ7srYE0U7EDpnJuxS6qVJVKpddee+2nf/qnv/rVr77w\nwgthZBkphFCv13//93//008/3dnZKZVK77333jjvNWbzLlSxG79Joyt2IYSVlZXLly+/9957\n3S9GvN0EP9cEl9rKdpA3iUdJIFfOvZP0nRDNBg0jy0hnPX1wPNyYzesdixYNodvd3T3rNfsG\n4Y1u0llfDz0SFeqiaba9Eykm/rlGHxznUk/w1kCaWccOSEKj0QghNJvNX//1Xx9xWrSa2tOn\nT6Nvoy+ieQYhhFdeeSWEsLGx0Ww2oyMHBwe9q7idZW1tLXpKs9nc2NgIIXzta18b+prdE6Lj\n5zZpUKvVOuuhW7duhRBu374dQvi5n/u57vGJf66hzr3UvS2c7VsD8zfvZAlk27l3kqhMNahv\nhFn0dXeFjqFndoaNIesd8XZW8/qe1VeNG72O3egm9Ta+m/aKxeJZFyc6Z3ChuMl+rote6sEW\nTvDWQJoJdsBUhiaMPru7u93A9OzZs76+zr5XODo66kaN6Py+V9vf3+8GlN3d3dERpPviUZ9j\nqVTqXe649zWj6RFDTxjRpN7GP3v2LGrYiGAXxcShXZ+T/Vx9Rl/qwRZO8NZAmpk8AeSZyQHA\nQjHGDgAgJwQ7AICcEOwAAHLCGDsAgJxQsQMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAg\nJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJC\nsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJ/5/\nOoTzN92PW9kAAAAASUVORK5CYII=",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Running knn for Minimum Validation Error k with prob=TRUE\n",
    "prediction2 <- knn(train_input, test_input, train_output, k=15, prob = TRUE)\n",
    "prob <- attr(prediction2, \"prob\")\n",
    "# Plot ROC and AUC for KNN\n",
    "library(ROCR)\n",
    "KNNPred <- prediction(prob, caravan_df_test$Number_of_mobile_home_policies)\n",
    "KNNPerf <- performance(KNNPred, \"tpr\", \"fpr\")\n",
    "plot(KNNPerf, colorize=TRUE)\n",
    "abline(a=0, b=1, lty=2, lwd=3, col=\"black\")\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$KNNU[i] <- prob[i]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "An object of class \"performance\"\n",
       "Slot \"x.name\":\n",
       "[1] \"None\"\n",
       "\n",
       "Slot \"y.name\":\n",
       "[1] \"Area under the ROC curve\"\n",
       "\n",
       "Slot \"alpha.name\":\n",
       "[1] \"none\"\n",
       "\n",
       "Slot \"x.values\":\n",
       "list()\n",
       "\n",
       "Slot \"y.values\":\n",
       "[[1]]\n",
       "[1] 0.5304873\n",
       "\n",
       "\n",
       "Slot \"alpha.values\":\n",
       "list()\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#AUC\n",
    "performance(KNNPred, \"auc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<dl class=dl-horizontal>\n",
       "\t<dt>acc</dt>\n",
       "\t\t<dd>0.566</dd>\n",
       "\t<dt>sens</dt>\n",
       "\t\t<dd>0.642857142857143</dd>\n",
       "\t<dt>spec</dt>\n",
       "\t\t<dd>0.56113769271664</dd>\n",
       "\t<dt>ppv</dt>\n",
       "\t\t<dd>0.0848115299334812</dd>\n",
       "\t<dt>npv</dt>\n",
       "\t\t<dd>0.961293260473588</dd>\n",
       "\t<dt>lor</dt>\n",
       "\t\t<dd>0.83356726883031</dd>\n",
       "</dl>\n"
      ],
      "text/latex": [
       "\\begin{description*}\n",
       "\\item[acc] 0.566\n",
       "\\item[sens] 0.642857142857143\n",
       "\\item[spec] 0.56113769271664\n",
       "\\item[ppv] 0.0848115299334812\n",
       "\\item[npv] 0.961293260473588\n",
       "\\item[lor] 0.83356726883031\n",
       "\\end{description*}\n"
      ],
      "text/markdown": [
       "acc\n",
       ":   0.566sens\n",
       ":   0.642857142857143spec\n",
       ":   0.56113769271664ppv\n",
       ":   0.0848115299334812npv\n",
       ":   0.961293260473588lor\n",
       ":   0.83356726883031\n",
       "\n"
      ],
      "text/plain": [
       "       acc       sens       spec        ppv        npv        lor \n",
       "0.56600000 0.64285714 0.56113769 0.08481153 0.96129326 0.83356727 \n",
       "attr(,\"negative\")\n",
       "[1] \"Not Purchased\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Corresponding Performance Measures\n",
    "KNNPrediction <- factor(as.factor(prediction2), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "KNNActual <- factor(as.factor(caravan_df_test_normalized[,'Number_of_mobile_home_policies']), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "library(crossval)\n",
    "CMNB <- confusionMatrix(KNNActual, KNNPrediction, negative = \"Not Purchased\" )\n",
    "diagnosticErrors(CMNB)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### KNN Using Oversampled Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 254,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "'Ratio of 1s to 0s- 1:15.7298850574713'"
      ],
      "text/latex": [
       "'Ratio of 1s to 0s- 1:15.7298850574713'"
      ],
      "text/markdown": [
       "'Ratio of 1s to 0s- 1:15.7298850574713'"
      ],
      "text/plain": [
       "[1] \"Ratio of 1s to 0s- 1:15.7298850574713\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#First let's recall the ratio of 1s and 0s in our original training dataset\n",
    "paste0(\"Ratio of 1s to 0s- 1:\", as.numeric(table(caravan_df_train$Number_of_mobile_home_policies))[1]/as.numeric(table(caravan_df_train$Number_of_mobile_home_policies))[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead><tr><th scope=col>Number_of_mobile_home_policies</th><th scope=col>Count</th></tr></thead>\n",
       "<tbody>\n",
       "\t<tr><td>0   </td><td>5474</td></tr>\n",
       "\t<tr><td>1   </td><td>2436</td></tr>\n",
       "</tbody>\n",
       "</table>\n"
      ],
      "text/latex": [
       "\\begin{tabular}{r|ll}\n",
       " Number\\_of\\_mobile\\_home\\_policies & Count\\\\\n",
       "\\hline\n",
       "\t 0    & 5474\\\\\n",
       "\t 1    & 2436\\\\\n",
       "\\end{tabular}\n"
      ],
      "text/markdown": [
       "\n",
       "Number_of_mobile_home_policies | Count | \n",
       "|---|---|\n",
       "| 0    | 5474 | \n",
       "| 1    | 2436 | \n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "  Number_of_mobile_home_policies Count\n",
       "1 0                              5474 \n",
       "2 1                              2436 "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Now let's duplicate the observations with dependent variable = 1 to make the ratio approximately 1:2\n",
    "caravan_df_train_KNN_OverDummy <- caravan_df_train[caravan_df_train$Number_of_mobile_home_policies==1,]\n",
    "caravan_df_train_KNN_Over <- NULL\n",
    "for (i in 1:7){\n",
    "    caravan_df_train_KNN_Over <- rbind(caravan_df_train_KNN_Over, caravan_df_train_KNN_OverDummy)\n",
    "}\n",
    "caravan_df_train_KNN_Over <- rbind(caravan_df_train_KNN_Over, caravan_df_train[caravan_df_train$Number_of_mobile_home_policies==0,])\n",
    "\n",
    "#Let's verify the number of 1s and 0s in our oversampled data\n",
    "oversampled.Frequency.Number_of_mobile_home_policies <- data.frame(Number_of_mobile_home_policies = levels(as.factor(caravan_df_train_KNN_Over$Number_of_mobile_home_policies)), Count = as.numeric(table(caravan_df_train_KNN_Over$Number_of_mobile_home_policies)))\n",
    "oversampled.Frequency.Number_of_mobile_home_policies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "'Ratio of 1s to 0s After Oversampling- 1:2.24712643678161'"
      ],
      "text/latex": [
       "'Ratio of 1s to 0s After Oversampling- 1:2.24712643678161'"
      ],
      "text/markdown": [
       "'Ratio of 1s to 0s After Oversampling- 1:2.24712643678161'"
      ],
      "text/plain": [
       "[1] \"Ratio of 1s to 0s After Oversampling- 1:2.24712643678161\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Let's verify the ratio of 1s to 0s after over sampling\n",
    "paste0(\"Ratio of 1s to 0s After Oversampling- 1:\", as.numeric(table(caravan_df_train_KNN_Over$Number_of_mobile_home_policies))[1]/as.numeric(table(caravan_df_train_KNN_Over$Number_of_mobile_home_policies))[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 258,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAMAAADKOT/pAAAAM1BMVEUAAAAAAP9NTU1oaGh8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\ni9u51zISAQAAAABJRU5ErkJggg==",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Classify Using KNN\n",
    "#Normalizing each variable\n",
    "fun <- function(x)\n",
    "{    \n",
    "    a <- mean(x)    \n",
    "    b <- sd(x)    \n",
    "    (x - a)/(b)\n",
    "}\n",
    "library(class)\n",
    "\n",
    "caravan_df_train_normalized <- apply(caravan_df_train_KNN_Over, 2, fun)\n",
    "caravan_df_test_normalized <- apply(caravan_df_test, 2, fun)\n",
    "\n",
    "train_input <- as.matrix(caravan_df_train_normalized[,-86])\n",
    "train_output <- as.vector(caravan_df_train_normalized[,86])\n",
    "test_input <- as.matrix(caravan_df_test_normalized[,-86])\n",
    "\n",
    "kmax <- 15\n",
    "ER1 <- rep(0,kmax)\n",
    "ER2 <- rep(0,kmax)\n",
    "\n",
    "for (i in 1:kmax){\n",
    "    prediction <- knn(train_input, train_input,train_output, k=i)\n",
    "    prediction2 <- knn(train_input, test_input, train_output, k=i)\n",
    "\n",
    "#Confusion Matrix For Test Data\n",
    "CM2 <- table(prediction2, caravan_df_test_normalized[,'Number_of_mobile_home_policies'])\n",
    "\n",
    "#Error Rate on the test sample\n",
    "ER2[i] <- (CM2[1,2]+CM2[2,1])/sum(CM2)\n",
    "}\n",
    "#Plotting the %Error on the test sample\n",
    "plot(c(1,kmax),c(0.15,0.8),type=\"n\", xlab=\"k\",ylab=\"Error Rate\")\n",
    "lines(ER2, col=\"blue\")\n",
    "legend(12, 0.75, c(\"Test\"),lty=c(1,1), col=c(\"blue\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 259,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Minimum Validation Error k: 1"
     ]
    }
   ],
   "source": [
    "# Minimum Validation Error k\n",
    "z <- which.min(ER2)\n",
    "cat(\"Minimum Validation Error k:\", z)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "      predicted\n",
       "actual    0    1\n",
       "     0 3513  249\n",
       "     1  196   42"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Running knn for Minimum Validation Error k\n",
    "prediction2 <- knn(train_input, test_input, train_output, k=z)\n",
    "actual <- ifelse(caravan_df_test_normalized[,'Number_of_mobile_home_policies']>0,1,0)\n",
    "predicted <- ifelse(as.numeric(prediction2)==1,0,1)\n",
    "table(actual, predicted)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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\nrkJggg==",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Running knn for Minimum Validation Error k with prob=TRUE\n",
    "prediction2 <- knn(train_input, test_input, train_output, k=15, prob = TRUE)\n",
    "prob <- attr(prediction2, \"prob\")\n",
    "# Plot ROC and AUC for KNN\n",
    "library(ROCR)\n",
    "KNNPred <- prediction(prob, caravan_df_test$Number_of_mobile_home_policies)\n",
    "KNNPerf <- performance(KNNPred, \"tpr\", \"fpr\")\n",
    "plot(KNNPerf, colorize=TRUE)\n",
    "abline(a=0, b=1, lty=2, lwd=3, col=\"black\")\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$KNNO[i] <- prob[i]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "An object of class \"performance\"\n",
       "Slot \"x.name\":\n",
       "[1] \"None\"\n",
       "\n",
       "Slot \"y.name\":\n",
       "[1] \"Area under the ROC curve\"\n",
       "\n",
       "Slot \"alpha.name\":\n",
       "[1] \"none\"\n",
       "\n",
       "Slot \"x.values\":\n",
       "list()\n",
       "\n",
       "Slot \"y.values\":\n",
       "[[1]]\n",
       "[1] 0.4286675\n",
       "\n",
       "\n",
       "Slot \"alpha.values\":\n",
       "list()\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#AUC\n",
    "performance(KNNPred, \"auc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 270,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<dl class=dl-horizontal>\n",
       "\t<dt>acc</dt>\n",
       "\t\t<dd>0.83675</dd>\n",
       "\t<dt>sens</dt>\n",
       "\t\t<dd>0.30672268907563</dd>\n",
       "\t<dt>spec</dt>\n",
       "\t\t<dd>0.870281765018607</dd>\n",
       "\t<dt>ppv</dt>\n",
       "\t\t<dd>0.130124777183601</dd>\n",
       "\t<dt>npv</dt>\n",
       "\t\t<dd>0.952020936318697</dd>\n",
       "\t<dt>lor</dt>\n",
       "\t\t<dd>1.08796631932574</dd>\n",
       "</dl>\n"
      ],
      "text/latex": [
       "\\begin{description*}\n",
       "\\item[acc] 0.83675\n",
       "\\item[sens] 0.30672268907563\n",
       "\\item[spec] 0.870281765018607\n",
       "\\item[ppv] 0.130124777183601\n",
       "\\item[npv] 0.952020936318697\n",
       "\\item[lor] 1.08796631932574\n",
       "\\end{description*}\n"
      ],
      "text/markdown": [
       "acc\n",
       ":   0.83675sens\n",
       ":   0.30672268907563spec\n",
       ":   0.870281765018607ppv\n",
       ":   0.130124777183601npv\n",
       ":   0.952020936318697lor\n",
       ":   1.08796631932574\n",
       "\n"
      ],
      "text/plain": [
       "      acc      sens      spec       ppv       npv       lor \n",
       "0.8367500 0.3067227 0.8702818 0.1301248 0.9520209 1.0879663 \n",
       "attr(,\"negative\")\n",
       "[1] \"Not Purchased\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Corresponding Performance Measures\n",
    "KNNPrediction <- factor(as.factor(prediction2), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "KNNActual <- factor(as.factor(caravan_df_test_normalized[,'Number_of_mobile_home_policies']), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "library(crossval)\n",
    "CMNB <- confusionMatrix(KNNActual, KNNPrediction, negative = \"Not Purchased\" )\n",
    "diagnosticErrors(CMNB)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### 5. Classification using Naive Bayes - Unbalanced Data, Undersampled Data, Oversampled Data, SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#NB Pre-Processing\n",
    "caravan_df_trainNB <- caravan_df_train\n",
    "caravan_df_testNB <- caravan_df_test\n",
    "\n",
    "caravan_df_trainNB$Customer_Subtype <- as.factor(caravan_df_trainNB$Customer_Subtype)\n",
    "caravan_df_testNB$Customer_Subtype <- as.factor(caravan_df_testNB$Customer_Subtype)\n",
    "caravan_df_trainNB$Customer_main_type <- as.factor(caravan_df_trainNB$Customer_main_type)\n",
    "caravan_df_testNB$Customer_main_type <- as.factor(caravan_df_testNB$Customer_main_type)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message:\n",
      "\"package 'e1071' was built under R version 3.3.3\""
     ]
    },
    {
     "data": {
      "text/plain": [
       "\n",
       "Naive Bayes Classifier for Discrete Predictors\n",
       "\n",
       "Call:\n",
       "naiveBayes.default(x = X, y = Y, laplace = laplace)\n",
       "\n",
       "A-priori probabilities:\n",
       "Y\n",
       "         0          1 \n",
       "0.94022673 0.05977327 \n",
       "\n",
       "Conditional probabilities:\n",
       "   Customer_Subtype\n",
       "Y              1            2            3            4            5\n",
       "  0 0.0202776763 0.0138838144 0.0409207161 0.0091340884 0.0078553160\n",
       "  1 0.0373563218 0.0172413793 0.0718390805 0.0057471264 0.0057471264\n",
       "   Customer_Subtype\n",
       "Y              6            7            8            9           10\n",
       "  0 0.0195469492 0.0074899525 0.0526123493 0.0485933504 0.0284983559\n",
       "  1 0.0344827586 0.0086206897 0.1465517241 0.0344827586 0.0258620690\n",
       "   Customer_Subtype\n",
       "Y             11           12           13           15           16\n",
       "  0 0.0263061746 0.0173547680 0.0303251735 0.0009134088 0.0029229083\n",
       "  1 0.0258620690 0.0459770115 0.0373563218 0.0000000000 0.0000000000\n",
       "   Customer_Subtype\n",
       "Y             17           18           19           20           21\n",
       "  0 0.0016441359 0.0034709536 0.0005480453 0.0042016807 0.0027402265\n",
       "  1 0.0000000000 0.0000000000 0.0000000000 0.0057471264 0.0000000000\n",
       "   Customer_Subtype\n",
       "Y             22           23           24           25           26\n",
       "  0 0.0171720862 0.0451223968 0.0319693095 0.0146145415 0.0085860431\n",
       "  1 0.0114942529 0.0114942529 0.0143678161 0.0057471264 0.0028735632\n",
       "   Customer_Subtype\n",
       "Y             27           28           29           30           31\n",
       "  0 0.0089514066 0.0045670442 0.0153452685 0.0208257216 0.0363536719\n",
       "  1 0.0028735632 0.0000000000 0.0057471264 0.0114942529 0.0172413793\n",
       "   Customer_Subtype\n",
       "Y             32           33           34           35           36\n",
       "  0 0.0242966752 0.1395688710 0.0316039459 0.0376324443 0.0381804896\n",
       "  1 0.0229885057 0.1321839080 0.0258620690 0.0229885057 0.0459770115\n",
       "   Customer_Subtype\n",
       "Y             37           38           39           40           41\n",
       "  0 0.0222871757 0.0577274388 0.0564486664 0.0129704056 0.0365363537\n",
       "  1 0.0287356322 0.0660919540 0.0545977011 0.0000000000 0.0143678161\n",
       "\n",
       "   Number_of_houses\n",
       "Y       [,1]      [,2]\n",
       "  0 1.111619 0.4119539\n",
       "  1 1.094828 0.2933985\n",
       "\n",
       "   Avg_size_household\n",
       "Y       [,1]      [,2]\n",
       "  0 2.671721 0.7915461\n",
       "  1 2.790230 0.7547864\n",
       "\n",
       "   Avg_age\n",
       "Y       [,1]      [,2]\n",
       "  0 2.990318 0.8167002\n",
       "  1 3.005747 0.7816124\n",
       "\n",
       "   Customer_main_type\n",
       "Y             1           2           3           4           5           6\n",
       "  0 0.092071611 0.079649251 0.151077822 0.009499452 0.101205700 0.036719035\n",
       "  1 0.137931034 0.189655172 0.169540230 0.000000000 0.043103448 0.011494253\n",
       "   Customer_main_type\n",
       "Y             7           8           9          10\n",
       "  0 0.096821337 0.269272927 0.114176105 0.049506759\n",
       "  1 0.057471264 0.255747126 0.120689655 0.014367816\n",
       "\n",
       "   Roman_catholic\n",
       "Y        [,1]     [,2]\n",
       "  0 0.6949214 1.009413\n",
       "  1 0.7212644 0.901428\n",
       "\n",
       "   Protestant\n",
       "Y       [,1]     [,2]\n",
       "  0 4.612715 1.719874\n",
       "  1 4.850575 1.637308\n",
       "\n",
       "   Other_religion\n",
       "Y       [,1]     [,2]\n",
       "  0 1.068871 1.015955\n",
       "  1 1.086207 1.042904\n",
       "\n",
       "   No_religion\n",
       "Y       [,1]     [,2]\n",
       "  0 3.275301 1.593205\n",
       "  1 2.994253 1.645874\n",
       "\n",
       "   Married\n",
       "Y       [,1]    [,2]\n",
       "  0 6.149616 1.92118\n",
       "  1 6.715517 1.62725\n",
       "\n",
       "   Living_together\n",
       "Y        [,1]      [,2]\n",
       "  0 0.8916697 0.9736863\n",
       "  1 0.7557471 0.8251076\n",
       "\n",
       "   Other_relation\n",
       "Y       [,1]     [,2]\n",
       "  0 2.317501 1.733062\n",
       "  1 1.864943 1.488333\n",
       "\n",
       "   Singles\n",
       "Y       [,1]     [,2]\n",
       "  0 1.911765 1.810822\n",
       "  1 1.508621 1.573350\n",
       "\n",
       "   Household_without_children\n",
       "Y       [,1]     [,2]\n",
       "  0 3.227073 1.616579\n",
       "  1 3.281609 1.674011\n",
       "\n",
       "   Household_with_children\n",
       "Y       [,1]     [,2]\n",
       "  0 4.283887 2.003963\n",
       "  1 4.557471 2.011387\n",
       "\n",
       "   High_level_education\n",
       "Y       [,1]     [,2]\n",
       "  0 1.426379 1.599997\n",
       "  1 2.005747 1.868893\n",
       "\n",
       "   Medium_level_education\n",
       "Y       [,1]     [,2]\n",
       "  0 3.331933 1.758583\n",
       "  1 3.655172 1.774475\n",
       "\n",
       "   Lower_level_education\n",
       "Y       [,1]     [,2]\n",
       "  0 4.624954 2.288806\n",
       "  1 3.747126 2.290028\n",
       "\n",
       "   High_status\n",
       "Y       [,1]     [,2]\n",
       "  0 1.865546 1.777452\n",
       "  1 2.359195 2.047069\n",
       "\n",
       "   Entrepreneur\n",
       "Y        [,1]      [,2]\n",
       "  0 0.3936792 0.7725679\n",
       "  1 0.4655172 0.8116341\n",
       "\n",
       "   Farmer\n",
       "Y       [,1]      [,2]\n",
       "  0 0.536719 1.0731614\n",
       "  1 0.295977 0.7212588\n",
       "\n",
       "   Middle_management\n",
       "Y       [,1]     [,2]\n",
       "  0 2.876872 1.824324\n",
       "  1 3.247126 2.037648\n",
       "\n",
       "   Skilled_labourers\n",
       "Y       [,1]     [,2]\n",
       "  0 2.238400 1.729670\n",
       "  1 1.928161 1.725553\n",
       "\n",
       "   Unskilled_labourers\n",
       "Y       [,1]     [,2]\n",
       "  0 2.329741 1.695094\n",
       "  1 1.939655 1.615605\n",
       "\n",
       "   Social_class_A\n",
       "Y       [,1]     [,2]\n",
       "  0 1.593168 1.707187\n",
       "  1 2.054598 1.904010\n",
       "\n",
       "   Social_class_B1\n",
       "Y       [,1]     [,2]\n",
       "  0 1.597187 1.328323\n",
       "  1 1.758621 1.359908\n",
       "\n",
       "   Social_class_B2\n",
       "Y       [,1]     [,2]\n",
       "  0 2.200950 1.531862\n",
       "  1 2.227011 1.490668\n",
       "\n",
       "   Social_class_C\n",
       "Y       [,1]     [,2]\n",
       "  0 3.779320 1.927586\n",
       "  1 3.433908 2.032502\n",
       "\n",
       "   Social_class_D\n",
       "Y        [,1]     [,2]\n",
       "  0 1.0880526 1.314966\n",
       "  1 0.7413793 1.050215\n",
       "\n",
       "   Rented_house\n",
       "Y       [,1]     [,2]\n",
       "  0 4.298502 3.083770\n",
       "  1 3.267241 3.017789\n",
       "\n",
       "   Home_owners\n",
       "Y       [,1]     [,2]\n",
       "  0 4.710632 3.084522\n",
       "  1 5.732759 3.017789\n",
       "\n",
       "   X1_car\n",
       "Y       [,1]     [,2]\n",
       "  0 6.012970 1.559355\n",
       "  1 6.471264 1.377791\n",
       "\n",
       "   X2_cars\n",
       "Y       [,1]     [,2]\n",
       "  0 1.314030 1.202742\n",
       "  1 1.353448 1.209384\n",
       "\n",
       "   No_car\n",
       "Y       [,1]     [,2]\n",
       "  0 1.990318 1.608292\n",
       "  1 1.474138 1.371559\n",
       "\n",
       "   National_Health_Service\n",
       "Y       [,1]     [,2]\n",
       "  0 6.306175 1.966783\n",
       "  1 5.818966 2.107756\n",
       "\n",
       "   Private_health_insurance\n",
       "Y       [,1]     [,2]\n",
       "  0 2.700219 1.970320\n",
       "  1 3.181034 2.107756\n",
       "\n",
       "   Income_._30\n",
       "Y       [,1]     [,2]\n",
       "  0 2.615638 2.097796\n",
       "  1 1.913793 1.768613\n",
       "\n",
       "   Income_30.45.000\n",
       "Y       [,1]     [,2]\n",
       "  0 3.536536 1.880069\n",
       "  1 3.528736 1.925644\n",
       "\n",
       "   Income_45.75.000\n",
       "Y       [,1]     [,2]\n",
       "  0 2.703325 1.930749\n",
       "  1 3.172414 1.826522\n",
       "\n",
       "   Income_75.122.000\n",
       "Y        [,1]     [,2]\n",
       "  0 0.7791377 1.151583\n",
       "  1 1.0632184 1.300199\n",
       "\n",
       "   Income_.123.000\n",
       "Y        [,1]      [,2]\n",
       "  0 0.2029594 0.5562646\n",
       "  1 0.1982759 0.4720394\n",
       "\n",
       "   Average_income\n",
       "Y       [,1]     [,2]\n",
       "  0 3.754476 1.313607\n",
       "  1 4.255747 1.295346\n",
       "\n",
       "   Purchasing_power\n",
       "Y       [,1]     [,2]\n",
       "  0 4.187797 1.997301\n",
       "  1 5.000000 2.010061\n",
       "\n",
       "   private_third_party_insurance\n",
       "Y        [,1]      [,2]\n",
       "  0 0.7478992 0.9519903\n",
       "  1 1.1379310 0.9889582\n",
       "\n",
       "   third_party_insurance_firms\n",
       "Y         [,1]      [,2]\n",
       "  0 0.04018999 0.3655047\n",
       "  1 0.03735632 0.3153808\n",
       "\n",
       "   third_party_insurane_agriculture\n",
       "Y         [,1]      [,2]\n",
       "  0 0.07435148 0.5095003\n",
       "  1 0.02873563 0.3116963\n",
       "\n",
       "   car_policies\n",
       "Y       [,1]     [,2]\n",
       "  0 2.859335 2.914510\n",
       "  1 4.718391 2.421328\n",
       "\n",
       "   delivery_van_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.04914140 0.5358994\n",
       "  1 0.03448276 0.4542029\n",
       "\n",
       "   Contribution_motorcycle.scooter\n",
       "Y        [,1]      [,2]\n",
       "  0 0.1740957 0.8967414\n",
       "  1 0.1954023 0.9058132\n",
       "\n",
       "   lorry_policies\n",
       "Y        [,1]      [,2]\n",
       "  0 0.0100475 0.2523224\n",
       "  1 0.0000000 0.0000000\n",
       "\n",
       "   trailer_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.02027768 0.2106038\n",
       "  1 0.03160920 0.2439605\n",
       "\n",
       "   tractor_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.09444647 0.6066462\n",
       "  1 0.06321839 0.5437865\n",
       "\n",
       "   agricultural_machines_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01388381 0.2360467\n",
       "  1 0.00000000 0.0000000\n",
       "\n",
       "   moped_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.22415053 0.8288564\n",
       "  1 0.07183908 0.4837534\n",
       "\n",
       "   life_insurances\n",
       "Y        [,1]      [,2]\n",
       "  0 0.1899890 0.8880731\n",
       "  1 0.2701149 1.0417913\n",
       "\n",
       "   private_accident_insurance_policies\n",
       "Y          [,1]      [,2]\n",
       "  0 0.014249178 0.2141054\n",
       "  1 0.005747126 0.1072113\n",
       "\n",
       "   family_accidents_insurance_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01333577 0.1772903\n",
       "  1 0.04597701 0.3531022\n",
       "\n",
       "   disability_insurance_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.02064304 0.3516211\n",
       "  1 0.06896552 0.6404808\n",
       "\n",
       "   fire_policies\n",
       "Y       [,1]     [,2]\n",
       "  0 1.782974 1.875056\n",
       "  1 2.531609 1.806690\n",
       "\n",
       "   surfboard_policies\n",
       "Y           [,1]       [,2]\n",
       "  0 0.0007307271 0.04273894\n",
       "  1 0.0028735632 0.05360563\n",
       "\n",
       "   boat_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01260504 0.2221641\n",
       "  1 0.11781609 0.6793995\n",
       "\n",
       "   bicycle_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.02411399 0.1534171\n",
       "  1 0.04310345 0.2033824\n",
       "\n",
       "   property_insurance_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01534527 0.2062034\n",
       "  1 0.02011494 0.1769025\n",
       "\n",
       "   social_security_insurance_policies\n",
       "Y        [,1]      [,2]\n",
       "  0 0.0411034 0.3817870\n",
       "  1 0.1494253 0.7045138\n",
       "\n",
       "   Number_of_private_third_party_insurance\n",
       "Y        [,1]      [,2]\n",
       "  0 0.3918524 0.4904490\n",
       "  1 0.5775862 0.4946549\n",
       "\n",
       "   Number_of_third_party_insurance_firms\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01479722 0.1350379\n",
       "  1 0.01436782 0.1191730\n",
       "\n",
       "   Number_of_third_party_insurane_agriculture\n",
       "Y         [,1]       [,2]\n",
       "  0 0.02137377 0.14464007\n",
       "  1 0.00862069 0.09257971\n",
       "\n",
       "   Number_of_car_policies\n",
       "Y       [,1]      [,2]\n",
       "  0 0.540190 0.6005142\n",
       "  1 0.908046 0.5657299\n",
       "\n",
       "   Number_of_delivery_van_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.010778224 0.13269204\n",
       "  1 0.005747126 0.07570049\n",
       "\n",
       "   Number_of_motorcycle.scooter_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.04055535 0.2289949\n",
       "  1 0.04885057 0.2288270\n",
       "\n",
       "   Number_of_lorry_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.002374863 0.06478277\n",
       "  1 0.000000000 0.00000000\n",
       "\n",
       "   Number_of_trailer_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01223968 0.1254846\n",
       "  1 0.01724138 0.1303570\n",
       "\n",
       "   Number_of_tractor_policies\n",
       "Y         [,1]     [,2]\n",
       "  0 0.03470954 0.245331\n",
       "  1 0.01724138 0.150853\n",
       "\n",
       "   Number_of_agricultural_machines_policies\n",
       "Y          [,1]      [,2]\n",
       "  0 0.006576544 0.1280668\n",
       "  1 0.000000000 0.0000000\n",
       "\n",
       "   Number_of_moped_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.07343807 0.2705058\n",
       "  1 0.02298851 0.1500825\n",
       "\n",
       "   Number_of_life_insurances\n",
       "Y         [,1]      [,2]\n",
       "  0 0.07307271 0.3628991\n",
       "  1 0.13218391 0.5576803\n",
       "\n",
       "   Number_of_private_accident_insurance_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.005480453 0.07383369\n",
       "  1 0.002873563 0.05360563\n",
       "\n",
       "   Number_of_family_accidents_insurance_policies\n",
       "Y          [,1]      [,2]\n",
       "  0 0.005845817 0.0762411\n",
       "  1 0.017241379 0.1303570\n",
       "\n",
       "   Number_of_disability_insurance_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.004201681 0.07514328\n",
       "  1 0.011494253 0.10674680\n",
       "\n",
       "   Number_of_fire_policies\n",
       "Y        [,1]      [,2]\n",
       "  0 0.5613811 0.5647991\n",
       "  1 0.7068966 0.4981333\n",
       "\n",
       "   Number_of_surfboard_policies\n",
       "Y           [,1]       [,2]\n",
       "  0 0.0003653635 0.01911274\n",
       "  1 0.0028735632 0.05360563\n",
       "\n",
       "   Number_of_boat_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.003836317 0.06471266\n",
       "  1 0.040229885 0.21091800\n",
       "\n",
       "   Number_of_bicycle_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.02995981 0.2027973\n",
       "  1 0.06034483 0.3117893\n",
       "\n",
       "   Number_of_property_insurance_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.007489953 0.08832124\n",
       "  1 0.014367816 0.11917295\n",
       "\n",
       "   Number_of_social_security_insurance_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01223968 0.1116132\n",
       "  1 0.04597701 0.2097368\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Classification Using Naive Bayes\n",
    "library(e1071)\n",
    "# Can handle both categorical and numeric input,\n",
    "# but output must be categorical\n",
    "caravan_df_trainNB$Number_of_mobile_home_policies <- factor(caravan_df_trainNB$Number_of_mobile_home_policies, labels = c(0,1))\n",
    "model <- naiveBayes(as.factor(Number_of_mobile_home_policies)~., data=caravan_df_trainNB)\n",
    "model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 273,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "      predicted\n",
       "actual    0    1\n",
       "     0  660 3102\n",
       "     1   16  222"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#NB Prediction\n",
    "tst <- caravan_df_testNB[,-86]\n",
    "prediction <- predict(model, newdata = tst)\n",
    "table(caravan_df_testNB$Number_of_mobile_home_policies, prediction, dnn=list('actual','predicted'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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\nEYIdAEAiBDsAgEQIdgAAifj/AfAq51Po0tjeAAAAAElFTkSuQmCC",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC and AUC for KNN\n",
    "probs <- predict(model, tst, type=\"raw\")\n",
    "library(ROCR)\n",
    "NBPred <- prediction(probs[,2], caravan_df_testNB$Number_of_mobile_home_policies)\n",
    "NBPerf <- performance(NBPred, \"tpr\", \"fpr\")\n",
    "plot(NBPerf, colorize=TRUE)\n",
    "abline(a=0, b=1, lty=2, lwd=3, col=\"black\")\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$NB[i] <- probs[i]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 276,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "An object of class \"performance\"\n",
       "Slot \"x.name\":\n",
       "[1] \"None\"\n",
       "\n",
       "Slot \"y.name\":\n",
       "[1] \"Area under the ROC curve\"\n",
       "\n",
       "Slot \"alpha.name\":\n",
       "[1] \"none\"\n",
       "\n",
       "Slot \"x.values\":\n",
       "list()\n",
       "\n",
       "Slot \"y.values\":\n",
       "[[1]]\n",
       "[1] 0.6759942\n",
       "\n",
       "\n",
       "Slot \"alpha.values\":\n",
       "list()\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#AUC\n",
    "performance(NBPred, \"auc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 324,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<dl class=dl-horizontal>\n",
       "\t<dt>acc</dt>\n",
       "\t\t<dd>0.2205</dd>\n",
       "\t<dt>sens</dt>\n",
       "\t\t<dd>0.932773109243697</dd>\n",
       "\t<dt>spec</dt>\n",
       "\t\t<dd>0.175438596491228</dd>\n",
       "\t<dt>ppv</dt>\n",
       "\t\t<dd>0.0667870036101083</dd>\n",
       "\t<dt>npv</dt>\n",
       "\t\t<dd>0.976331360946746</dd>\n",
       "\t<dt>lor</dt>\n",
       "\t\t<dd>1.08252615091648</dd>\n",
       "</dl>\n"
      ],
      "text/latex": [
       "\\begin{description*}\n",
       "\\item[acc] 0.2205\n",
       "\\item[sens] 0.932773109243697\n",
       "\\item[spec] 0.175438596491228\n",
       "\\item[ppv] 0.0667870036101083\n",
       "\\item[npv] 0.976331360946746\n",
       "\\item[lor] 1.08252615091648\n",
       "\\end{description*}\n"
      ],
      "text/markdown": [
       "acc\n",
       ":   0.2205sens\n",
       ":   0.932773109243697spec\n",
       ":   0.175438596491228ppv\n",
       ":   0.0667870036101083npv\n",
       ":   0.976331360946746lor\n",
       ":   1.08252615091648\n",
       "\n"
      ],
      "text/plain": [
       "      acc      sens      spec       ppv       npv       lor \n",
       "0.2205000 0.9327731 0.1754386 0.0667870 0.9763314 1.0825262 \n",
       "attr(,\"negative\")\n",
       "[1] \"Not Purchased\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Corresponding Performance Measures\n",
    "NBPrediction <- factor(as.factor(prediction), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "NBActual <- factor(as.factor(caravan_df_testNB$Number_of_mobile_home_policies), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "library(crossval)\n",
    "CMNB <- confusionMatrix(NBActual, NBPrediction, negative = \"Not Purchased\" )\n",
    "diagnosticErrors(CMNB)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Naive Bayes Using Under Sampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead><tr><th scope=col>Number_of_mobile_home_policies</th><th scope=col>Count</th></tr></thead>\n",
       "<tbody>\n",
       "\t<tr><td>0  </td><td>348</td></tr>\n",
       "\t<tr><td>1  </td><td>348</td></tr>\n",
       "</tbody>\n",
       "</table>\n"
      ],
      "text/latex": [
       "\\begin{tabular}{r|ll}\n",
       " Number\\_of\\_mobile\\_home\\_policies & Count\\\\\n",
       "\\hline\n",
       "\t 0   & 348\\\\\n",
       "\t 1   & 348\\\\\n",
       "\\end{tabular}\n"
      ],
      "text/markdown": [
       "\n",
       "Number_of_mobile_home_policies | Count | \n",
       "|---|---|\n",
       "| 0   | 348 | \n",
       "| 1   | 348 | \n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "  Number_of_mobile_home_policies Count\n",
       "1 0                              348  \n",
       "2 1                              348  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Under Sampling Data\n",
    "#Taking all the observations with dependent variable = 1\n",
    "caravan_df_train_NB_Under <- caravan_df_trainNB[caravan_df_trainNB$Number_of_mobile_home_policies==1,]\n",
    "\n",
    "#Randomly select observations with dependent variable = 0\n",
    "zeroObs <- caravan_df_trainNB[caravan_df_trainNB$Number_of_mobile_home_policies==0,]\n",
    "set.seed(123457)\n",
    "rearrangedZeroObs <-  zeroObs[sample(nrow(zeroObs), length(caravan_df_train_NB_Under$Customer_Subtype)),]\n",
    "\n",
    "#Appending rows of randomly selected 0s in our undersampled data frame\n",
    "caravan_df_train_NB_Under <- rbind(caravan_df_train_NB_Under, rearrangedZeroObs)\n",
    "\n",
    "#Let's verify that number of 1s and 0s in our undersampled data are equal\n",
    "undersampled.Frequency.Number_of_mobile_home_policies <- data.frame(Number_of_mobile_home_policies = levels(as.factor(caravan_df_train_NB_Under$Number_of_mobile_home_policies)), Count = as.numeric(table(caravan_df_train_NB_Under$Number_of_mobile_home_policies)))\n",
    "undersampled.Frequency.Number_of_mobile_home_policies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 279,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\n",
       "Naive Bayes Classifier for Discrete Predictors\n",
       "\n",
       "Call:\n",
       "naiveBayes.default(x = X, y = Y, laplace = laplace)\n",
       "\n",
       "A-priori probabilities:\n",
       "Y\n",
       "  0   1 \n",
       "0.5 0.5 \n",
       "\n",
       "Conditional probabilities:\n",
       "   car_policies\n",
       "Y       [,1]     [,2]\n",
       "  0 2.841954 2.911919\n",
       "  1 4.718391 2.421328\n",
       "\n",
       "   Income_._30\n",
       "Y       [,1]     [,2]\n",
       "  0 2.577586 2.032095\n",
       "  1 1.913793 1.768613\n",
       "\n",
       "   moped_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.20402299 0.8007820\n",
       "  1 0.07183908 0.4837534\n",
       "\n",
       "   fire_policies\n",
       "Y       [,1]     [,2]\n",
       "  0 1.887931 1.883211\n",
       "  1 2.531609 1.806690\n",
       "\n",
       "   Lower_level_education\n",
       "Y       [,1]     [,2]\n",
       "  0 4.589080 2.363409\n",
       "  1 3.747126 2.290028\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Running NB on Undersampled Data\n",
    "library(e1071)\n",
    "# Can handle both categorical and numeric input,\n",
    "# but output must be categorical\n",
    "NBModelUnder <- naiveBayes(as.factor(Number_of_mobile_home_policies)~car_policies+Income_._30+moped_policies+fire_policies+Lower_level_education, data=caravan_df_train_NB_Under)\n",
    "NBModelUnder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 280,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "      predicted\n",
       "actual    0    1\n",
       "     0 2040 1722\n",
       "     1   71  167"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#NB Prediction for undersampling\n",
    "tst <- caravan_df_testNB[,-86]\n",
    "prediction <- predict(NBModelUnder, newdata = tst)\n",
    "table(caravan_df_testNB$Number_of_mobile_home_policies, prediction, dnn=list('actual','predicted'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 281,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94\nAAAgAElEQVR4nOzdf2yc+X0f+O+TKKgDo+AGSeR6lZManMFt4AaSt9dYm22qVkqRyulwEdi7\nkRQIRhBpQ6EsGkM8ZLMhL3DJ1pu74d22XtQstb2iFbJkrbTFcpAIbi0avsCmjMJrzhnbQnNb\n9MjDYstp0pK92t1e1Mz98VjjWQ45HM7P5/vM6wViwXnmB7/7cPTMm5/vr6RWqwUAAOL3PcNu\nAAAAvSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAH\nAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAA\nOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJAT\ngh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHY\nAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0A\nQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDk\nhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4I\ndgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAH\nAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAAOSHYAQDkhGAHAJATgh0AQE4IdgAA\nOSHYAQDkhGAHAJATgh0AQE4IdgAAOXFs2A2IQ5Ikw24CAJAhtVpt2E3Yh2DXrmz+/gCAgcl+\noUdXLADA4bKf6oJgBwBwqMZUl+VOPMEOAKCVWFJdEOwAAFrIQqpbW1ubmJhIkuTGjRtra2st\nHinYAQDsb/Cpbmtra2FhIUmSiYmJNMOVSqULFy6USqUQwuLi4oULF2ZnZw9scMYrihmRJE4U\nAIyiNNvtiQF9CgZbW1unTp1qPLK8vHz58uWlpaXnnntubGysWq3evn17enr63r1758+fb34F\nFTsAgAPVarWBFXe+8IUvhBCWl5fTH5qmukKhcP369bGxsRDC8ePHr127FkK4c+fOvq+gENUW\nFTsAoK5PwaC5OpgkSbFYvHnzZuuH1cW0QHGlUnnw4EGlUpmenm48XiwWT5w48eSTT46Pjw+r\nbQAAPbe+vv7tb3+7+XihUNj38XEEu93d3RdeeGFxcXHfe+s5b2ZmZm5uboDtAgDomaWlpeef\nf/7WrVvpiLoQwtmzZxsfsLu7+/nPfz6EcOXKlX1fIY4expWVlbSP+cUXX3z88cdPnjzZeG+1\nWn3nnXdmZ2dLpdLy8vKlS5d63gBdsQBAXZ+Cwe7u7tWrV9MJsPu+ftoJWygUVldX929YFHll\nYmKiVCptbm7uiXSN0okkLf5XuyHYAUBetRiy1uIp/QsGKysrr7322r55ZmJi4sqVKy1qWHHk\nlTbPeAe/mPYbEMWJAgCOpLOV6jIbDOJY7iQdIbi1tdXiMem9B40lBADYIwu7SvRWHMEuHSE4\nNTVVLper1eqee6vVarlcnpqaCgePJQQAaJTZVLe1tbWysjI7O5u8140bNxYWFu7fv9/iuRkt\nJDabnZ2dn59v/ZjJycmXXnopnUXSW5mtuAIAHegy1fUvGJRKpYmJidaPaV7Zri6mvFKpVN54\n44233367cR27QqFw7ty58fHxJ554on/r2Al2AJAb3dfq+hQMyuXymTNnlpaWnnnmmePHj4cQ\nKpXK7du3H3vssZs3b25tbb311lt37txZXFzc2Ng4ffr0Pg2TV9oh2AFAPvSkB7ZPwSDtn9zZ\n2WnsfkzX/VhdXU0nElQqlSeeeOKgop280hbBDgByoFfj6ga2pVj9eOOCbi2WAYlj8gQAQJcy\nO1uiLq3J7e7uNh5Mb6arFh8qV8EunTMy7FYAAJmT/VQXHi3uUSwWK5VKeqRarb766qshhKWl\npRDC7u7urVu36jeb5aqHsbMFitvMgnk6UQCQYwd/sPdsI4P+hYd9lwEpFAq3b98eGxurbyn2\n6quvprMr9jZMXmmHMXYAEIskCf3+0O5rMFhbW/vSl76UxrvJycmPfexj9f0X0i3FLl68eNDi\nbvJKWwQ7AMiaFlWzHn9of9/en5Q8zGhX3rFhNwAAoEOZDFfDJNgBALTUHJceDqEV7RDsAIA4\nWPoi1WK2qGAHAESjdd9rZ+tj5Ekcwe5Iq9ON8q8TACLS2wpcPS30fspjxuJSi/+7OBYovnfv\n3rCbAAD0Xq12tK+DRLH+8ADEEezOnz+/ublZKBRmZmZqhxl2YwGAgep7qjvW9JVVMS3PtrW1\nderUqeXl5UuXLg34R1vHDgB6ricrCQ+iVvcDTevY7fSxLlipVB48eFCpVKanpxuPF4vFEydO\nPPnkk+Pj4wc9N7K8sra2duHChc3NzZMnTw7y5wp2ANBz3Qe7AfXA/nBTsPv9vvy43d3dF154\nYXFxsfXDZmZm5ubm9r0rjq7YuvPnz9dqtQGnOgCgh5LkO19dv07extXdvXt3cXGxUCisr69v\nbm7uGWy2vb29sbFRKBTm5+dXVlb2fQWFqLao2AFAr0TTA1v3waaK3b/ryw+dmJgolUqteybT\nkWmFQmF1dbX5XnmlLYIdAPRKND2wdYMKdm2uw9fiYZF1xQIA1OWs7FIoFEIIW1tbLR6T3ps+\nsplgBwDEpF6uG1yqG9RyJ1euXAkhTE1NlcvlarW6595qtVoul6empuqP3LelAADRqNVqeR0i\ndenSpTfffHN+fr5UKrV42OTk5MWLF/e9K5/npefy+gYCgMHryeSJgfrRpjF2/3d/17F74403\n3n777cZ17AqFwrlz58bHx5944on8rGM3LIIdAPRE2o8a2YfqYINdN3TFAgADlclE1FI8ccnk\nCQCAnIgnggIAcep+k4khiycuxdNSACBSx0LtD4/2jDaX6mUPXbEAQLbUV6pLoq/1DZqKHQDQ\nZ0eJG4PeLqwd8cSleFoKAGRb8pFH370bwsMQ3n301f4rZDDVRUWwAwB65Fio/cvOn53dVBdP\nXDLGDgAYvuymuqjEE0EBgIxJ/tKj79K+145fJ+OpLp64FE9LAYDsqf2Lbl8h66kuKoIdANCJ\n5GIvXiSKVBdPXDLGDgDoUJflujhSXVQEOwDgyJKLoXa32xephzmprlfiqS0CAAOR/LVHMyHq\n8yHSb7qbIbGvOCJdPHEpnpYCAINxLNQ+O+w20BHBDgCgpXjikjF2AAA5EU8EBQD6LPm13o+i\ny4N44lI8LQUA+in5tVD7TB9eNklCLJMk4ifYAQD9TXXpNxFnu3jiUjwtBQB6KlkI4T+H8DCE\nh/1NdUHFblAEOwAYRclCqN3s5+tLdcMg2AHAqEj+XggPQ3i3/z8oZ6kunrgUT0sBgK7V/mrf\nf0TeUl1UBDsAGAnJ3wu1X+7/T8llqosnLsXTUgDgiJLfatr1ta8/LpepLiqCHQDkU/JbofYL\nA/xxOU518cSleFoKABwmef07y5cM+ufmONVFRbADgJxIfifUnhnOj67VanneYSKeuBRPSwFg\nxCTlR981FuEaFytp/H7Ye7zmM9LFRrADgKw6FmofHnYbiIpgBwDQUjxxKZ6WAkB+JZvv7W8d\ndr8qkRLsAGDIks1QOzXsRtBCPHEpnpYCQC4kuw2LBkdVmcvzvNe8EOwAYHCS3VAbG3YjOlJf\nqS5JkpHLdvHEpe8ZdgMAYFTkINUFFbtsiyeCAkDMpLqIxROXVOwAoO+kOgYjnggKAJFIvvXe\n2w+lOgZEsAOAnkkjXe39w25HL0h13xVPXIqnpQCQYcl/C+HdnES6INVFS7ADgF743lrt/cnh\nD4uBVLdXPHHJ5AkA6FYSarUg1TF88URQAMiAJPy/j7499ugrV2q1mh0m9ornlxxPSwFgqNJI\nVwt/fNgN6TuRLl6CHQDsLwnvhPC+EEL6cTkKkY79xROX4mkpAAxQEt6phQ8OuxVwNIIdALxH\nEt4JIUh1xEiwA4DvUqhjH/HEJcudAMB3jE6qS5KkcVkTciOeCAoAvZaE/6fh1rHRSXX1b0yA\nbUs8cSmelgJA7z2shR8ddhsGyvrD+SbYATCikvBvpTraEk9ciqelANCpJPzrEMJ7P/VG7hNQ\nqhsFI/e2BmAkPayFHx92G4ZJqutKPHEpnpYCwBEl4RuPvh3pzzupbnSM9BsdgLx7WAt/dtht\nGDKpbqQIdgDkUxL+pVQn1fVGPHHJAsUAkE9S3QiKJ4ICQNuU60IItVotzXZSXbfiiUvxtBQA\n2pCE9RCCD7iUSDdqvO8ByIkk/F4IoRZ+atgNIXfiiUvxtBQADpaE3xPpQLADIHpJ+FIt/MVh\nt4L8iicumRULQNykOqgT7AAgYkmSNC5rwoiLp7YIAE1GvFxXj3RJkpgA20fxxKV4WgrACEjC\nnYbPpj0fUsfe+/2xEIJUl5LqSAl2AGRFEu7UwrPDbkUcpLqBiicuxdNSAHIqCcshBB9J7ZPq\nOIh/RQAMUxKWa+HysFsRE6luCOKJS2bFAkA0pDpaiyeCAhC/JNxuuJXOflCua5dUNzTxxKV4\nWgpAHrxbC9eH3YYoSXW0Q1csAAOShFvDbkKspDpCe4tRq9gBMDAPa+HGsNsQpVqtln6iS3XD\nEU9ciqelAMQsCZ+T6roh0o2CdnaHqz9m37eErlgAgJaONX31x4MHDwqFQjevINgB0HfKddCO\n8fHx27dvF4vFEMLMzMz29natQfqYPTf3iGbP4Gq1+vrrrz///POFQuH69etpnl1bW7tz587i\n4mKhULhy5cqlS5f69NNtrgywRxLmHn17rOG/4b0HQ8OmroId0fqdvT2kyV/pb+f4/fv3n3rq\nqRDC+vr62bNnv/ND2xhnGUde2d3dvXr1aqlUqh+5d+/et771rYmJicaHzczMzM3NNT27BwQ7\ngD2SMFcLs8NuBQzEwINdCKFarX72s5+dn58vFovXrl0bGxtrJ9jF0RV79+7dUqm0vLyc1h6X\nl5dffvnliYmJpaWlnZ2dWq22s7OzvLw8Pz+/srIy7MYCQCfaWcyC4RjUGLtGx48fn5ubW11d\nnZ6evnr1arlcbudZcRSiJiYmSqVSY1PTt/7Ozs7Y2FjjwUKhsLq62vMGqNgBNFKu6zkr1WXa\nF5oqdn95cL+mra2tz3zmM4uLi+nNPHTFNtce961G9m+NH8EOoJFg11tSXdYNNdilbt269fzz\nz4d8dMW2M/V3d3e3zUcC0A2prrekuggMoyt2j+vXr7eYDFsXR7C7cuVKCKE+fm5lZSUNcCsr\nK2meCyHcvXu3/kgA+iEJv5GE35Dqekiqo1GSJBMTE5VKpfNXiOJtZFYswAAk4VdDCC0qErXw\n6UG2J/ekumh8uakr9i/0a+hX+k19JuyRXyGWd5J17AD6Kgm/Wgu/OexWjBCpLiaDDXYbGxuz\ns7PpeiAXL148UryTV9oi2AG5J9gNklQXma80Bbs/18dgV6vVdnd3X3311enp6RDC0tLST/zE\nT5w+fbqtV/B+aodgB+RVEv5qCMdCeJ9UNzBSXXwGHuzSm1tbW3fu3EnjXQihWCx+5CMf+dCH\nPjQ2NnZQGU9eaYtgB+RSEn6pFv7+sFsxWqS6KH2tKdidHUSwS21tbX31q1997bXXGicbND/s\nO6+Qp3eVdewA2peEXwohCHaD179PK/plqMGubmtr66233vrGN77x1ltvLS4uCnatnnWoPJ0o\ngBBCEj5ZC/9w2K2AGHy9Kdj9D209r7NM0k3kGMYSe33T2Ylo51k27wMAGmWz4pOrYAdAm5Tr\nIIO6D4uCHcDIkergaOKJS/G0NIRKpfLgwYNKpVKf95sqFosnTpx48sknx8fHh9U2gAxKwif3\nPS7VQV7FMXlid3f3hRdeWFxcbP0wW4oB1CnLDZF5r3nzZtPkiT89zN9vizfY9wy8MZ24e/du\num/Y+vr65uZm7b22t7c3NjYKhcL8/PzKysqwGwswZEn4+ST8wrBbMbrq8+1MvGPw4ihETUxM\nlEqlzc3NkydPHvSYra2tU6dOFQqF1dXVnjdAxQ6ISBJ+vhb+8bBbMaKsP5xPD5oqdn8qo7/f\nOPJKmzVtCxQDSHVDJNXlVjzBLo6u2EKhEELY2tpq8Zj03vSRAKMpCT8/7CaMLqkuz441fWVV\nHMHuypUrIYSpqalyuVytVvfcW61Wy+Xy1NRU/ZEAoyYJH0/Cx2vhHyvXDYVURw9VKpVSqbSw\nsJC818LCwsrKSqVSafHcaHoYZ2dn5+fnWz9mcnLypZdeGhsb6/lP1xULZE0SPt54sxb+ybBa\nglSXf/+mqSv2Q335XXe/DEhMeaVSqbzxxhtvv/124zp2hULh3Llz4+PjTzzxRP/WsRPsgKxJ\nwseFuSyQ6kbCZlOw+5N9+XWvrKxcvny5UCi8+OKLjz/++J45o9Vq9Z133pmdnS2VSsvLy5cu\nXWp+BXmlLYIdkClSXUZIdaNiUMGu+2VA5JW2CHZApgh2WSDVjZC3m4Ldj/RrFY7Q3TIgcUye\nAKBOqsuI+seqVEevdL8MiGAHAB1KN0Aadivov0Etd9L9MiB6GNuiKxbICOU6GILtpq7YP9Gv\nSm2Xy4DIK20R7IB+S8K5R98ea/hvo/elBwU7GLQ/aAp2P9THLvhulgGRV9oi2AH9loSna+Er\nw24FsJ/BBrtuGGMHAJATGd7tDGA0JOGjIQQX5Gxqc/kJci6ef53xtBQgd9JIVwtfG3ZD2F99\npToDcoiFYAcwHEn4qEiXZdYf5rviiUvG2AEMy8NhN4ADSXVEKp4ICgADIdWxVzxxKZ6WAkQu\nCR8O4X3DbgWHkOqImmAHMAhJ+HAtvDnsVnAIqY79xROXjLEDgBCkOnJBsAPoO+W67JPqyId4\naosAEUrCEyEEF9uISHU0q33vsFvQNtcagP6qhQfDbgKHqJfrpDpiJ9gB9EsSnpDqolCr1ewt\nQQsRrTlpjB0AqNWREyp2AH2hXAe5EVHFTrAD6JkknAjh2KNLqwssMGiuOwA99LAW3h52G4Ae\ni6hiZ4wdQG8k4QO1sD3sVgAjTcUOoCtJ+MCwm0Bb0jVNTJIg3wQ7gG4p1GVffaU6y5rQAV2x\nACMhCT847CZwONuFMTpU7AA6lIQfrIU/GHYrOIRUR/dU7ABg+KQ6Ro2KHUAnlOuyT6qjV1Ts\nAPJMqss+qY7RpGIHcDRSXfZJdfSWih0ADIdUxyhTsQNoJQnvf/Ttd3aAVa7LMqmOESfYAbT2\nsBb+67DbQLtqtZodJui5iLpiBTsAckWkY5QJdgAHSsIfU64DIqrYmTwBAJATKnYA3/H+kIQQ\n3g0hhPBH4XtcIYFURBU7ly2AEEJ4f0i+FYzNAuIm2AGj5VRIQsPf3w9DeDeqP8cJj9Y0MUmC\ngYnoEiHYASNnU2UuZvWV6pIkke1gD5MnAIiG9YehNRU7ILc+HJI9HSjvDqch9IZUx7DoigUY\nvmMhvKnXNS+kOmiHYAfk0+mQlKW6vJDqGK6IKnbG2AGQaVIdtE/FDoDskurIAhU7AOiWVAdH\npWIH5JABdjkg1ZEdKnYAQ3M6JIc/iMyrhzmpDton2AF5cywE5bp8qNVqUh0cia5YAIBWIuqK\nFeyAPPhoSEJUF1+AfhDsgOg9HZKv6XsF+iaiPxqNsQMAyAkVOyBuT4fkK8p10UrXNDFDgoxT\nsQOAQ9RXqmtcsg7ohoodAENg/WEiomIHMAj6YSMl1UGfCHYADJRUB/2jKxaI1TnlughJdcQo\noq5YwQ6Iw8827AD7MIR3h9gUOiXVQb8JdkAcjoXwuvpczKQ64hVRxc4YOyACz4REqouaVAeD\noWIHZN0zwSJncZPqiF1EFTvBDsi0n1eri1+tVrPDBAyGYAdk1M8r1OWISEfUVOwAunUshN9S\nqwM4CpMnAAByQsUOAKCViLpiVewAAHJCxQ7Iok+aOREb817JsYgqdoIdkDmfDMk/NG0iKvWV\n6pIkke1giAQ7YKBuhqT+t2/jH8EPH918GIJUFxfrD5N7KnYA+/jVkCwIbfki1UGmCHYAdEiq\nY0REVLEzKxYYkF8NyW8q1+WIVAcZJNgBcGRSHWSTrligX+ZCEkJ4N6peDNoh1TFqIrqICXZA\nX8yFZFbHax5JdZBlgh3Qrc+FJDxar+RhCO+GEEKQ6nJJqmM05bBiVy6Xv/jFL05PT4dH/5gX\nFhauXr16/PjxPrYOiMGxEK6LcaOhVqvZYQKyrK3JE7Ozs2fOnElTXd309PS1a9eq1Wp/GgbE\n4VZIpLqRUqvVpDpGzcOmr/7Z3d1dWVmZmJioHymVSjdu3EiSJEmS2dnZ+/fvt3j64Xu/rK2t\nXbhwYWZmZnp6+rHHHguP/lBLjy8tLV2/fr0X/yOZZpMcaPQPHvW9pgQ7IN8+07R79YtJX+rW\n1Wr12rVrpVIphFCr1XZ3d69evZrebDQzMzM3N7fvKxxesXv55ZdDCNPT02NjY43Hz58/H0J4\n/vnnO2s6EKl/EJJfDLVfDLXrj76G3SKA/hpYxe71118vlUrFYnFnZyeE8PnPf75UKi0tLW1v\nb6fF8u3t7aWlpfn5+ZWVlX1f4fBCVONwij1DK0ZnpIWKHaRuh+RhCL8ozAGjZK6pYvc/9adi\nlyarnZ2dtJq252Zqd3f3scceKxQKq6urza9weMVucnIyfZU9x7e2tur3AqPgdkiuhppUB9BX\n9RhXKBQabzbe29w/mzo82D377LMhhGKx2JjtyuXy1NRU/d4B6HIsIQBtSq+rw24FZMjAumKX\nlpbCo9pZCCGdxlC/mUrz2MzMzL6v0FYP48LCwp4psakWY/d6q/uxhF3SFQtpuW7YraDvrFQH\nzX6jqSv2b/SnK3Zra+vUqVOFQuHKlSs/+ZM/efLkyYWFhS9/+cuvvPLKyZMnQwjlcnlxcXFx\ncXFjY+P06dPNr9DWOnY3b9786Z/+6fo6diGEYrH49NNPnz17tof/My3UxxJeu3YtNIwlfOaZ\nZ9KF9KrV6uuvv/78889/+MMfvnTp0mBaBaNDqhsRUh3sa2ALFJ88eXJjY2N2dvby5cuNx/cU\ns1ZXV/dNdaHNit3QdT+WsPsGRHGioE8Eu1Eg1cFBfr2pYve3+lOxq7t///43v/nNN954Y3Fx\nsX6wWCyOj49/9KMfbbE9RBx5Zc/024mJiVKp1Nzy/s3SFewYZVLdKJDqoIVfbQp2/3Ofg13H\nDp880WIU7cTERONshv7pfiwh0JnlpssZ+SPVQW60taXYvqrVaqlUOmi2bW/9zM/8TAhhampq\nZWVla2urUCgUi8Wpqal6tiuXyy+88EII4ROf+MQA2gMjRbku36Q6ONQgtxRrtra2Vl8GZGFh\noVwut3jw/pMnyuXymTNnGo8cVLQbzDp23Y8lBKCZVAeZ0jyobHZ2dn5+vn4znca6vLx80FTR\n/St2p0+fbiexTU5OfupTnzpSizt2+vTp1dXV9fX1paWlPW0rFourq6vb29vpOn5A9+6EJP0a\ndkPoI6kOMm5lZWV+fn55eXlnZ6e+pdjy8vLly5fX1tb2fcrRthQbWSZPMGr+WUh+Tg9srkl1\n0L6/3vRX7t/p55Zie2aL7rsMyOTk5Oc+97l9XsG/53YIdoyIfxGSEMK7ITwMQbDLPX+3Q5uG\nFewO+kfa4h9v55MnQgiVSmVhYaGbVwAy5X0h/KVQK4SaVDcK0p6dYbcCIjCsyRPpqiD7Omj4\nWVs7T6ysrOyZtdDo5s2b7bzIAPjrE7rxeyH5KXkOYNgWFhbGx8ff//73/8RP/EQI4f79+/W9\nvnZ3d+/evRtCuHLlyr7PPbyHsXWqO2irsqHoLNi1ude1vEi+/V5IQgiCHUCzX27qil1qb2pZ\nrzLJns7ZQqFw+/btxoF3dYdX7F577bUQwvr6+tmzZ9M5t9vb2yGEz372s/Pz89///d9/pBb3\nVWfZq51ntRn+IEYiHUAH+lHxqdVq6Rq9b731VgjhG9/4Rni0xEmqUChcuXLl4sWL+6a6cNRZ\nsaVSaWJiIq3SbW1tnTp1amZmZm5urkf/O9ll8gR5pfsV4FC/1FSx+9/j3VKs0cmTJ0MIf/AH\nf1D/vnHRPAAypcWekEAuHR7sisViCOH+/fshhA9+8IMhhDt37oQQKpVKn9u2V6VSKZVKCwsL\nyXstLCysrKwMvj0QO+W6fKtHOtkORsfhPYz17cXSR+7Z2mJpaen69et9bWIIYXd394UXXlhc\nXGz9sP71C+uKJU/WQ1Kfqy/Y5ZX1h6GHPtnUFfuPhtoV29U6duleXvWbc3Nz9VVVlpeXB5Dq\nQgh3795dXFwsFArr6+ubm5u199re3t7Y2CgUCvPz8ysrKwNoD8RrPSRPhdpPPfoadnPoC6kO\nRlYchah0S43Nzc10YN++0skchUKhMYb2ioodObAekhDCU8Jc3kl10HO/0FSxey2rkycOzysL\nCwsnTpy4dOnSYBq0rzYXqOvfAsWCHbFLC3XDbgV9J9VBP0QU7A7vip2enm6xQPFgpPtmpCu7\nHCS996AdNmCUSXUjQqqDPhnWlmIdODzYTU5OhhB2d3f735gDpftmTE1NlcvlarW6595qtVou\nl6empsLBO2zAKHvfsBvAAEh1kBvdLANyeA/j7u7uq6++urOzc/Xq1fHx8Z62/Aj2zMbd1+Tk\n5EsvvXTQWszd0BVL1L4Rko+o2OWaVAd99fGmrth/2p+u2O6XAWl354kWBnYRqVQqb7zxxttv\nv71nb41z586Nj48/8cQT/cudgh0x+kbDlUiwyzGpDvptYMFuZWXl8uXLhULhxRdffPzxx/fM\nGa1Wq++8887s7GypVFpeXt53/kNMwW6IBDvikkY6YW509G/qGBAGGOy6XwZEXoBHYUwAACAA\nSURBVGmLYEdmffO9l5t0SK9IB9BDzzQFu9X+BLvulwE52l6xQNa8L4QfD7X610dCTaoDiFT3\ny4AIdgAArQxsuZPulwE51re2Ab3370NSv6ZkfC0lAI7q0qVLb7755vz8fKlUavGwycnJixcv\n7nuXoWNtMcaOjPj3IflhPa0Ag/UzTWPs/nk/d57oZhkQeaUtgh0D9h+bLiLhUX1OsBtN5r3C\nEA042HVDVyxkzn8MyQ9IbzSoLzvlj0wYiojGvQh2kCH/JSTvhiDV0cj6w0D72p0VWy6X63uW\npUcWFhaa52sAXZLqaCTVAUfSVlV/zz6t6VOSJCkUCq+++urx48f72MBs0P3BAPyXkHy/VEcD\nqQ4y4lzTGLv/I6tj7A6v2K2trc3Pz8/MzOzs7DQev3fvXqlUev311/vWNhgJfxiS9GvYDSFb\npDqgA4cXotJty3Z2dsbGxvZMyxqdWVoqdvRDGua+T5WOJlIdZMrTTX97fzWrFbvD80pjehPs\noFf+MCQiHfuS6iBrIgp2h3fFTk5OhhB2d3f3HE+3KkvvBY6kJtVxAKkOMmhgW4p17/Bg9+yz\nz4YQisViY7arb1WW3gtA96Q6oEtt9TAuLCw0bmpRNzMzMzc314dWZY6uWHqoFpJEuY4DjM4Q\nF4jIn2nqin0jq12x7eaVcrn8xS9+sR7visXi008/ffbs2X62LUMEO3pFqgOITg6D3YgT7OgV\nwQ4gOhEFu8PH2DVPmwA69F+lOoD45GryxGOPPTY7O1sulwfQGsizb1mCGID+amu5k/n5+TNn\nziRJsrKyUqlUBtAsyKc/plwHEJ+IKnZtDR2rVqtf+9rXfvd3f3dxcTGEMDk5+bGPfezP//k/\nPzY21v8WZoIxdvTAt5Lwfu8izHuF+Hy4aYzdv8rqGLuj5ZVqtbq2tvbaa6+VSqUQwszMzCc+\n8YnTp0/3rXlZIdjRA4IdVqqDOD3RFOwq+Qh2dWtraxcuXEi/z+b/WG8JdvSAYDfypDqIVETB\n7vAxdo2q1WqpVLpx40aa6iYnJ9fX1/vTMMiR3cTMCaQ6iFdEY+yOtfOgPWPsCoXC8vLy+fPn\njx8/3ufmQV6o1Y02qQ4YjMOD3Y0bN9I8F0JYWlo6d+7c+Ph4n1sFObKbhDEf5CNNqgMG5vCh\nY0mSzMzM/OzP/uzobCDWzBg7juwPHn2WHwuC3SiT6iAHTjWNsdvK6hi7wyt2Ozs7o7OsCfTM\nwxA+kMV/8wySVAcM2P7BrnGZpccee6z1S7hawV7biVSHVAe5keXZEnscbVYsAEcl1QEDs3/F\nrvEy5JIER7NtZRO+W65zCYUciKhi19ZyJ0C7dMISQgihVquZdAUM3uHBrsW2hhMTEyGE1dXV\nnjcLIHZSHeRGRBW7tpY7CftdoarV6gc+8IF978off3nTFuU6gDz6wablTv5DVpc72X/yRLlc\nTh5JjyRN0lQ3OTk5uMZClr0t1QEwZPsHu9OnT7eT2CYnJz/1qU/1ukkQoU0TJgByK6K9Yjvv\nih0pumJp5d8k4b/39gDIrbGmrtj/lNWu2MMnT2Sz3QBD5+9eGBFZLtHtcfjOE42Lp+/LRY2R\nplw3qhqHILsMAhlhHTuAI7NdGIyU6Ct2dp6AtijXjSSpDsgsFTuAI5DqYARFVLHbf7mTPdbW\n1m7cuJF+X61Wb9y4kSTJjRs3KpVKP9sG2aZcN3qkOiDjDh/zWy6Xz5w5Ex5dxSYmJkqlUv3e\nBw8ejI+P97WJWWBwNPsQ7EaMVAcj63ubljv5o6wud3J4xe63f/u3Qwjr6+shhK2trVKpNDk5\nWavV0iO3b9/udxMBhk6qg1H2R+F79nwNu0UHOtoCxVtbW6dOnVpeXr506VIYpTWcVOzYS7lu\nlEh1MOKS8L1Nh/4om1eDo02euHPnTgjhx37sx/rTGIDMkeqA/fLS/zeEVrTh8FrizMxMCGFr\na6tcLk9PTxcKhdOnT4cQ7t+/X78XIK/qYU6qA7LvCJMnUqurq4VCITz6K3ZjYyPNefmmK5a9\ndMUCjIwkvL/p0LezGQwOr9idPn363r17aZhbXl5OvwkhTE5Orq+vj0Kqg72kOgAySSGqLSp2\no+7ryXfGVxwL4X0hhCDYAYyOJIw1HfpP2QwGdp6Aw3w9CX8mi/96AWCPdhdiuX///uzsbPLI\nwsJCOnkCcutrSfh6ItUBEJG2ehhnZ2fn5+ebj8/MzMzNzfWhVZmjK3YUfS0JH/VLHxWjsyon\n0IEk/GDTof+QzSvG4RW7tbW1+fn5QqGwsbFRe2RjY6NQKMzPz6+trQ2glQD9U1+prnHJOoAY\nHR7s0kWJX3nllcYJsKdPn37llVfq90KufCUJX/MBPyqsPwy04VjTV0YdbUux9u/KGV2xI+Qr\nSXja73pUSHVAO5LwgaZD1WxeNA6PnJOTk4uLi1tbWydPnmw8vrW1FUKoL2sHkfmdR5/o9T+9\n0m+kupEh1QFty26Jbo/Du2KfffbZEMLU1FS5XK4fLJfLU1NTIYRf+ZVf6V/joL9+thZ+thZ+\nphYu1MKFWjhXk+pGh1QH5JJZsW3RFZsfrychPPrT62f9TkeUVAccSRJONR3ayubVo928cv/+\n/a985SvT09PpzWKx+PTTT589e7afbcsQwS4/Xk/CM36VI02qA44qh8FuxAl2OSHVjTypDuhA\nRMEumsGA0K3XrWAy6qQ6oFPR5KUjbCm2sLBQ31Ls1q1bjXMpIAIPg3LdiKuHOakOyKuuJk8U\ni8WbN2/2oVWZoys2ev8kCR/3GwSgE0l4oulQJZvB4PCK3crKykFbik1PT9tSDAAgIw4Pdq+9\n9lqwpRgAMLpsKZYvumKjpysWgE4l4cNNh/5VNoPB4RW75eXl8GgDsUbpkWKx2I9mQS9JdQB0\nJUcVuxBCqVS6devW3NxcvTe2UqlMT0+fPn3azhNkxW+9dzWTYw3/DUGwGx2j05MADEwSTjcd\n+j+zeZ1ptyu2Tdn8n+yeYBeB30rCL/gdjTor1QH9EFGwy24tEY5AqkOqA/oomrx0eENdH4Hs\nk+oAQkQJFA6kXDfypDqgz6LJS9E0FPbxW7Z/RaoD+C7Bjsip1Y02qQ4YiGjy0uHr2AFkk1QH\nsEc0CRSgkVQHDFA0eUnFjmiZMzHCpDqAfQl2QHzqYU6qA2jUbrArl8sLCwtJktT/UF5YWKhW\nq31rGEArtVpNqgMGJV97xc7Ozs7Pz9dvpk9JkqRQKLz66qvHjx/vYwOzwZZi2fIPkxBCOGZK\nLACDkIQLTYfWshkMDq/Yra2tzc/Pz8zM7OzsNB6/d+9eqVR6/fXX+9Y2aPL3k/D3k/DJWvhk\nTaoDYFAGVLFLkmRiYqJSqXT+CofmzYmJiVKptLOzMzY2lvbD1p+y52aOqdgN2d9OwvtCOBbC\nL/ktADBoSfiZpkP/vB/BoD7grVgsXrt2bWxs7KivcHjFrlQqhRA6eGnopV+uSXUADMlAx9ht\nbGx8+ctffuyxx1ZWVnZ3d4/03MOD3eTkZAih+XW3trbq9wL0XONsLYDRcfr06du3bxeLxcuX\nLz/22GO3bt0ql8ttPvfwYPfss8+GEIrFYmO2K5fLU1NT9XuhXxaS8Ld9tI+ieqST7YAMGPSs\n2LGxsZs3b25ubhaLxeeff/7MmTNJkiwsLKytrW1tbbUo47U1dGxhYWF6err5+MzMzNzcXFcN\n74UBDPUzxm5oFpJw05kfOdYfBjIlCc80HVrt3xi7Pa+8tbX11a9+9bXXXktHx9Xt24B280q5\nXP7iF79Yj3fFYvHpp58+e/Zshw3vKcEuzwS70SPVAVkz3GBXt7W19dZbb33jG9946623FhcX\nuwp2w3Wkvpg+negoTlQOCXYjRqoDMigJH2869E8HH+zaEceWYg8ePCgUCsNuBdBfUh1Al+II\nduPj4+n0kBDCzMzM9vZ2rUH6mD03yYMFo+ZHiFQHZNiAJk90n2QO72E8tBt0kJfg+/fvP/XU\nUyGE9fX1+gg/Y+xySz/syJDqgCxLws83Hfp8Ni9WcVTs6s6ePbu9vT0zM/PUU08tLCwcddU+\nYiLVjQypDsi8QS93ssfa2trExESSJDdu3FhbW2vxyMODXW0/Dx48mJmZKRaLg78KHz9+fG5u\nbnV1dXp6+urVq+0v2QdkkFQH0Ghra2thYSHdNDbNcKVS6cKFC+laJ4uLixcuXJidnT3o6Z33\nMO7u7j722GNLS0vXr1/v7BW6tLW19ZnPfGZxcTG9qSs2b1TsRsbo7DoNRCoJn2w69I/6cdXa\n2to6depU45Hl5eXLly8vLS0999xzY2Nj1Wr19u3b09PT9+7dO3/+fPMrdN4Vm+4e+/zzz3f8\nCl06efLk5z73uaWlpWE1gD6S6kaJaU8AqS984QshhOXl5fTCmKa6QqFw/fr1NHcdP3782rVr\nIYQ7d+7s+wqdF6LW1tYuXLhQKBRWV1c7bX80VOwGTbADIDMGVrFr7sFIkqRYLN68ebP1w+oO\nH/3XelbssPphAQAGZdCzJerW19e//e1vNx8/aH3fzrtiJycn7927l6l1g5MksV94HijXATCS\n0gFmt27dqq/7cfbs2caxdLu7u7du3QohXLlyZd9XyFUPY2dDsNvMgnk6UdlVX5FYsAMgM5Lw\ny02H2hrif9TwsLu7e/Xq1XQC7P5bwSZJCKHFQLjDS4sLCwsnTpy4dOnSkVo2FJ1lr3aepRA4\nOCJdHpn3CuRPP65pY2Njq6urKysrr7322r4PKBQKV65caZHK2t15YsSvyCZPDIhO2DyyUh0Q\nuyT81aZDfzebF7TDx9hNTk6GEOzxQN9JdXkk1QEM0uFdsS+99NKHPvShYrF49erV8fHxAbTp\nIJVK5cGDB5VKZXp6uvF4sVg8ceLEk08+OdzmAXtIdUBeDG1W7FG12xXbwgCu17u7uy+88EJ9\nk4mDzMzMzM3N9aMBumIHQcUuX6Q6IDeS8NebDv2dIV7ZWgyT63y5k0G6e/fu4uJioVBYX1/f\n3Nzcs3Ht9vb2xsZGoVCYn59fWVkZdmMBqQ5gOOIoRE1MTJRKpc3NzZMnTx70mHR7tT7thKFi\n13fKdTki1QE5k4SbTYf+12xe3/bPK1mbCdtme/rXbMGuv34zCccsdJITUh2QPxEFuzi6YtP9\nLba2tlo8Jr03Uzth0JbfTMKv1qS6fJDqgJw61vSVUXEEu3TfjKmpqXK5XK1W99xbrVbL5fLU\n1FQ4eIcNsuU3k+9+kRdSHUBPVCqVUqm0sLCQvNfCwsLKykqlUmnx3Di6YkMIs7Oz8/PzrR8z\nOTn50ksvjY2N9fyn64rtsb+ZhF93PnMog5cOgO4l4debDv2tflzrul8GJJpgF0KoVCpvvPHG\n22+/3biOXaFQOHfu3Pj4+BNPPNG/dewEu16S6gCIysCC3crKyuXLlwuFwosvvvj444/vmTNa\nrVbfeeed2dnZUqm0vLy878ZirYJdm0Yh8Qh2vSTYARCVJPxG06G/0Y9g0P0yIIJdWwS7npHq\nAIjNwIJd98uAtJo8UWtPF+1nZPzN5DtfAMABul8GJI5ZseTBr9e+8wUAkRnQcifdLwOS3YVY\nyA/drzmSzZlVAPlw6dKlN998c35+vlQqtXjY5OTkxYsX970rplmxQ2SMXVcEu7ywUh0wmpLw\nmaZDL/bvMtjNMiCCXVsEu85JdXkh1QEja8DBrhv7d8Vms63AsEh1wGiLZuiayRP02cNhN4Cu\nSXUAsYgmgRKZX0u+8+byFoucVAcQ0YdZNA0lJr+WhM9IAHkg1QHERbCjp/7HJARvq5yQ6gCi\n4xOYXvtfJIA8kOoAGkSTl0yeAPaS6gAiFU0CBQamVqtZzBKgQTR5KZqGEgeLm+SFSAcQI8EO\nAKC1aPKSMXb0zqeS8L8p8wDA0ESTQAEAhiSavKRiBwCQE4IdjKgkSRqXNQEgB6IpLZJ1BthF\npR7pkiQxARbgMNHkJRU7GDnWHwbIq2gSKFn0KR158ZHqAI4umrwUTUPJnL+WhM+KBZGR6gDy\nTbCDUSHVAXQqmrxkjB0dUa6LjVQHMAqiSaBAx6Q6gO5Ek5dU7Dg65bqoSHUAo0Ow4+geDrsB\ntE2qAxgp0ZQWgQ7UarU020l1AF2IJi9F01Cy4kYSPicixESkAxgdgh0AQGvR5CVj7AAAciKa\nBEom6IcFYBRFk5eiaShDdiMJIUh1AJBlgh3teRjCLakuo8x7BeizaPKSMXa04Xoi1WVWfaW6\nxiXrABhN0SRQhuMXkxC8TbLL+sMANPKJzcGuJuG2rJBdUh3AoESTl3TFQpSkOgCaRZNAGTTl\nugyT6gAGK5q8pGIHkZHqADhINAmUgVKuyyqpDmAYoslLKnYQDakOgNaiSaAMzlXLoWWdVAcw\nWNHkJRU79qMfNnvq5TqpDoCDCHYQhzTPSXUAtBBNaZH+uvyo+9U7IsOkOoAhiebTMZqG0l8P\nQ7gjNABA3AQ7AIDWoslLxtgBAORENAkUAGBIoslLKnaE8GxigF1GJEnSuAoxABxJNAkUcq8e\n6ZIkMQEWIEuiyUsqdiNPuS4bbBcGQPeiSaD0xc8l4Z/JEMMn1QHQE4LdqPq5JIQg1WWBVAeQ\nedHkJV2xI0yqywCpDoAeiiaBQv5IdQCRiCYvqdiNJEPrMkCqA6DnokmgkCdSHUBUoslLKnYw\naFIdAH0STQKF3KjVamm2k+oAIhFNXoqmoZAnIh0A/aArFgAgJ1TsAABaiyYvqdgBAORENAkU\nAGBIoslLKnbQF0mSNC5rAgADEE0ChYjUI12SJCbAAsQvmrykYgc9Zv1hAIYlmgRKz/yc/sE+\nkuoA8iiavBRNQ+mZhyGUBI6+kOoAGC5dsSPmonJdv0h1AAydit3oUa7rA6kOIM/+WzRlERU7\n6JZUB0BGqNhBV6Q6gPx7OOwGtE3FbpRcTMJdyaOXpDoAMkXFbmSYNtEHtVotzXZSHUCexVOx\nE+xGxrshfEn46D2RDoDsEOwAAFqKp2JnjB0AQE6o2I2Gv5johwWAKGxtbX31q19988035+fn\nG49PTk5+6EMfevrpp8+ePXvQcxMjhNqRJJGfKMEOADqV/EHToR/qVzAolUoTExOtH1MsFm/e\nvLnvXbpi4RBJkjQuawIAfVIulycmJpaWlra3t2u1Wq1We/DgwczMTLFYrNVqm5ub9+7dm5yc\nnJ6eLpfL+76CrlhopR7poq/aAtCxQU2e+O3f/u0QwnPPPTc2NpYeGR8fv379+qlTp8bHxwuF\nwsmTJ3/kR35kcXHxi1/84unTp5tfwWdVW6L/UNcV2xHrDwMQQki2mw79ib4Eg4PWRk2SpFAo\nrK6utn5YiKgrdnd399atW2mn2Ozs7L4VSF1m9JBUB8B3PGz66o9CoRBC2N3dbTyY3iyVSu28\nQhzBbnd39+rVq88//3x6c35+/syZMwsLC8NtFTkm1QEweFeuXAkhFIvFSqWSHqlWq6+++moI\nYWlpKTyqc9VvNotjjN3du3dLpdK9e/fOnz8fQtjd3f3617/+8ssv7+zszM3NDbt1macf9oik\nOgDeY1Bj7C5dupSucrJnoZNCofDcc8+FEB577LH05jPPPLPvK8QxdGxiYqJUKu1parVavXbt\n2okTJz796U8fP348tOxy7lKsY+z+4qOAIti1TaoDYI9ks+nQn+xjMFhbW/vSl76UZrvJycmP\nfexjaRdtCGFiYuLKlSsXL16sz67Y29QoProOSmy7u7vFYjGEkNbtBLv3+Kkk/F5sbR42qQ6A\nZgMOdt2IY4zd5ORkaBpLGEIYGxubm5v7/d///dnZ2WG0i1yR6gDY36AmT3QvjjF2zz777OLi\n4uc///nr16833/vpT3/6Ax/4wOBblWnKdUck1QGQAxktJDZbWVm5fPlyOOBDt1wunzlzJv1e\nV2wIgl0n+teVD0DUkgdNh/5URoNBHF2xIYRLly5tbm6mI+qanT59ent7+6Cpv6PlpxKprjPp\n5i3DbgUAdC6jeTNrYqrYSXUA0FPJm02H/nQfd55oU9w7TwAA5Nu9e/e6fAXBLl+U6wCg5wY1\nK/b8+fObm5uFQmFmZqZ2mH1fIY5ZsW0y+B0AiNrJkydfeeWVU6dOffjDH7506dJRn65iF5I2\nDLuN9IbfJgCd2K9i17/8cPLkyXv37l2+fHlra+uoz81Vxa6zWl07z5IGcqD+S4xpKgwAWdXX\nj5Lz58939voqdjligN3BrD8MwCjIVcUO9iXVAdCVDO8htkdMwa5SqTx48KBSqUxPTzceLxaL\nJ06cePLJJ8fHx4fVNjJLqgNgdMQx2Gh3d/eFF15YXFxs/bCZmZm5ubl+NCCOUVlPJWE9840c\nLKkOgO4lX2k69OeGGQxaLAMSxxi7u3fvLi4uFgqF9fX1zc3NPeu4bG9vb2xsFAqF+fn5lZWV\nYTeWrJDqABg1MRSiQpiYmCiVSpubmydPnjzoMVtbW6dOnSoUCqurqz1vQAQVu6eSEIKKXZ1U\nB0CvJF9uOvQXMhoMMtqsPdpcebh/CxRHEOz+bBL+ZbZbOEBSHQA9FFGwi6MrtlAohBBaL9OX\n3ps+klEm1QHQY4PaUqx7cQS7K1euhBCmpqbK5XK1Wt1zb7VaLZfLU1NT9UcyyuphTqoDIEaV\nSqVUKi0sLOzZx2JhYWFlZaVSqbR4bkYLic1mZ2fn5+dbP2ZycvKll14aGxvr+U/XFQsAIyu5\n13Top/sSDLpfBiTzeaVBpVJ544033n777cZ17AqFwrlz58bHx5944on+rWMn2AHAyEq+0HTo\nL/clGKysrFy+fLlQKLz44ouPP/74njmj1Wr1nXfemZ2dLZVKy8vLly5d2qepWc8r2SDYAcDI\nGliw634ZkMznlWwQ7ABgZCW/03Tor/QlGHS/DEgckyc4xEekOgCIXvfLgAh2RCmdHzTsVgAw\nGga13En3y4Ac61fToG/qkS6CLnIAaNulS5fefPPN+fn5UqnU4mGTk5MXL17c9y6fi23JeoD4\nSBK+keHm9ZT1hwEYsOSfNB36RB+DQTfLgGQ7r2SGYJcRUh0AgzfgYNcNY+yIhlQHAK0ZYxe5\nj4zKBAKpDoChyfDmsHsIdvEbgU5YqQ4A2iHYkXVSHQBDFk/Fzhg7Mk2qA4D2qdhFLp6/ITog\n1QGQCfF82qrYxezHk/DNPMedepiT6gCgHSp2ZJpIB8DwqdjRd3kv1wEARyXYAQDkhK7YOCnX\nAcDA6Iqlj358VHabAACORMUuQg9D+Nf5Kdela5qYJAFAdqnYQTvqK9U1LlkHAHRGxY6hsf4w\nAHFQsYPWpDoA6DkVO4ZAqgMgJip29MuPJbHPnJDqAKBPBLuoSHUAwMF0xTI4Uh0AUYqnK1aw\ni8GPJuFYCMfiXr5OqgOAfhPsIvF/xZ2EpDoAIhZPxc4YOwahHuakOgDoHxU7BkSkAyBWKnYA\nAAyYil3m/ahNVAFgqFTs6KV/qxMTADicYAcAkBO6YumZdE0TkyQAyBtdsYya+kp1jUvWAQCD\npGKXbT+aRDHAzvrDAOSZih2jQ6oDgIxQsaMrUh0A+adixyiQ6gAgU1Ts6JBUB8CoULEj36Q6\nAMggwY4jk+oAIJt0xXI0Uh0AI0dXLHlVD3NSHQBkjYodRybSATBaVOzogUi2nQAAMkLFDgCg\nJRU7AAAGTMUOAKAlFTuiliRJ47ImAEAUBLus+u+GlqvqkU62A4C46IrNsGFMibX+MADspSuW\nGEl1ABA1FbusenfQP1CqA4D9qdjRlR9Owr8faLSS6gAgB1TskOoAoCUVOzo32HKdVAcAuaFi\nN9KkOgA4nIodHRr46LqUVAcAOSDYZckPD3RB4Hq5TqoDgHzQFZsxAyzX1Wq1JEmkOgA4hK5Y\noiDVAUCeqNgBALSkYgcAwICp2GXGYGdOAADtiqdiJ9hlxsMQ/qMRbwBA5wS7nEvXNDFJAgA6\nF0/Fzhi7PKuvVNe4wwQAkFeCXW7ZLgwARo2u2HyS6gCgZ3TFMkRSHQCMJhW7bPiBno2Bk+oA\noMdU7DiyXqx1ItUBwChTscuAH0ikOgDILhU7BkyqAwBU7DLg3W5fQKoDgD5SsaNd39+DaRP1\nMCfVAcAoU7HLgP/SgzQm0gEAgh0AQEu6YmlXPO8VACDjVOwAAFqKpwqjYgcAkBMqdtFI1zQx\nSQIABk3Fjt6qr1TXuGQdAEAjFbsIWH8YAIZJxY5ekeoAgDYJdkP1fUn4w1ZZTaoDANoXU1ds\npVJ58OBBpVKZnp5uPF4sFk+cOPHkk0+Oj48Pq239INUBQCbE0xWbRJEYdnd3X3jhhcXFxdYP\nm5mZmZub60cDkqQ/J+rgip1UBwAZkXyw6dC/y2iCiqMr9u7du4uLi4VCYX19fXNzs/Ze29vb\nGxsbhUJhfn5+ZWVl2I3tAakOADLkYdNXVmU0b+4xMTFRKpU2NzdPnjx50GO2trZOnTpVKBRW\nV1d73oC+VOwOKNdJdQCQKckPNx36/YwmqIw2a4821+bt3xK+Awt2Uh0AZE3yA02HdjKaoOLo\nii0UCiGEra2tFo9J700fGa/6uySbbxcAIMviCHZXrlwJIUxNTZXL5Wq1uufearVaLpenpqbq\nj4xaOnBw2K0AAB4xxq7nZmdn5+fnWz9mcnLypZdeGhsb6/lP731X7GEr2AEAGZH88aZD/zmj\nCSqjzdpXpVJ544033n777cZ17AqFwrlz58bHx5944on+rWMn2AHAyBLs8kawA4CRlXx/06F3\n+5igutmRQbBri2AHACNrYMGu+x0Z4pg8kTNJkiQZHncJALzHoCZPdL8jQ64qdlGsY2elOgCI\nS/J9TYce9iVBdb8jg2D3nqTVQk9OlFQHANHZLyn0JTx0vyNDrrpiO1sBrtaGnjRPqgOA3OhH\nfuh+R4ZcBbssk+oAgNa635EhV12x/dPlGDupDgDypC+byIcQut6RIaZg182yLl3q5vcn1QFA\nzvQv2IXudmSII9h1v6xLlzr+/Ul1AJA/fQ123YhjjF33y7oMhVQHAAxSeVm3ZQAADC9JREFU\nRvPmHt0v69KlDoK5VAcAeZXZil1Gm7VH98u6dN+ADl62f+0BAIYos8Eujq7Y7pd1GYoeroEH\nAHCoOIJd98u6AADkXkYLic26XNalS5mtuAIAg5fZYJDRZu2rm2VdupTZ3x8AMHiZDQYZbVbW\nZPb3BwAMXmaDQRxj7DIrSZLGZU0AAIZIsOtcPdLJdgBAFgh2HbL+MACQNYJdJ6Q6ACCDBLsj\nk+oAgGwS7I5GqgMAMkuwOwKpDgDIMsGuXVIdAJBxgt2RSXUAQDYdG3YD4mPVOgAgmzK6IUaO\nZXYTkhHh/A+X8z9czv9wOf/DNSLnX1csAEBOCHYAADkh2AEA5IRgBwCQE4IdAEBOCHYAADkh\n2AEA5IRgBwCQE4IdAEBOCHYAADkh2AEA5MRI7JsGADAKVOwAAHJCsAMAyAnBDgAgJwQ7AICc\nEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnBDgAgJwQ7AICcEOwAAHJCsAMAyAnB\nDgAgJwS7HltZWZmYmEiSZGJiYmVlZXd3tx9P4SAdnMy1tbUbN244/z3RzZu5UqkkSdK/to2C\nDs7/1tbWrVu3kiRJkuTWrVvVanUA7cyro57/3d1d1/+eW1tba/9Kks/zX6N3ZmZm9pzeycnJ\nnj+Fg3RwMpeXl/c8pVAo7OzsDKbBOdPNm3lnZ6dQKLgidaOD87+xseH93ytHPf/b29vpe37P\n+d/e3h5Ym/Nnc3Oz/WyT189fl9GeefDgQfrPcnNzs1arbW5upv9oHzx40MOncJAOTmZ6CVha\nWqp/kt27dy89MqBG50iXb+Zisdj+5ZhmnZ3/QqFQf8rOzk76W1heXh5Qo3Okg/O/tLS052yn\nf2e6/nSsftrbuZLk+PPXZbRn0n+TGxsb9SPr6+shhNXV1R4+hYN0cDLTj7E9B8WLznTzZk4f\n6cx3o4Pznz6gMVjs7OykH3X9bWsedXD+933D+1fQsfTP8noR7tDH5/jz1xi7nnn77bdDCB/8\n4AfrRx5//PEQQqVS6eFTOEgHJ/PmzZu1R1dSutTxm7larT711FPNfSIcSQfn/5vf/GYI4Sd/\n8ifrR8bGxmq12urqah8bmlMdnP/mftjWx2ntwoULy8vLc3NzbT4+z5+/w06W+bHv+Wx9kjt4\nCgfpyclMKxYzMzM9bdpI6Pj8p5Fue3vbO78bHZz/ycnJ9N7V1dU0TDQOS+BIOjj/aYBu7orN\nQcVoKNIe1Vrbl/0cf/4mNRWLHkmn4ew5n/se7OYpHKQnJ3Ntbe3ChQsbGxunT5/ueQvzrbPz\nXyqVJiYm1tfXz549653fjY6vP8VicXp6un6wUCio2HWgs/f/2trayy+/XCqV0puFQuFXfuVX\nzp8/38+W5l+bV5Icf/7qioXvqFarL7/8crFYlOoGY2tra2Jiolgsnj17dthtGV1f/vKX65Mn\nlpaWSqVSPWfQb9/61rcaz3apVPrWt741xPaQD4IdhBDC7u7utWvXTp8+ffPmzWG3ZVR85jOf\nKRQK165dG3ZDRtorr7xy8uTJEMLY2NgzzzwTQvjd3/3dYTdqJKytrU1MTNS7YtNgPTExsba2\nNuymETfBrmc6GPFqkGwPdXkyi8XiiRMn2h94yx5HPf+3bt1aXFycm5sbGxvrU5NGSsfv/zTV\npY4fPx5CWFxc7E2bRkkH5//OnTshhIsXL6Y3x8bGnnvuufpx+i3Hn7+CXc+cO3cuhNC4bnv6\nfX2Brp48hYN0fDKr1eqNGzdCCJ/+9Kf72cCcO+r5f/7550MIZ86cSR5Jjzd+T/s6eP+7zvRQ\nB+c/DdCNf9ik3wvWg5Hjz1/BrmfGx8dDCO+88079SPr9iRMnevgUDtLZySyXyx/4wAd+6Id+\naG5uLi1X0Blv5uHq+PpTLpfrR9IPNkvPdKCD899cMUr3s8pxJSlT8nzJGuQU3Hyz88RwdXAy\n0yU2LG7SE92/mV2RutHx+39ycjJd4iQd4xVCWF9fH1y786KD858ubrK8vFxfYqZ+ZECNzqk2\nryQ5/vx1Ge2ldF2oRntCQ/Mb7tCn0L6jnv/0Y8wfPL3Swfu//Xs5VAfnv3llE9efjnVw/puL\no85/9w66kjQfz+vnr8toL+3s7CwvL6epv1AoNP/h1fzGOvQptO+o578pzn3XYBueEx28/9u/\nl0N1dv43NjbSeOH606XOzr/rf8+1H+zy+vlrgWIAgJwweQIAICcEOwCAnBDsAAByQrADAMgJ\nwQ4AICcEOwCAnBDsAAByQrADAMgJwQ4AICcEOwCAnBDsAAByQrADAMgJwQ4AICcEOwCAnBDs\nAAByQrADAMgJwQ4AICcEOwCAnBDsAAByQrADAMgJwQ4AICcEOwCAnBDsAAByQrADAMgJwQ4A\nICcEOwCAnBDsAAByQrADAMgJwQ4AICcEOwCAnBDsAAByQrADAMgJwQ7oSnKwI71CXxvZP3sa\nXy6XW9ybBXtaCOSMYAfQGzdu3Dhz5sywW9FK9lsIdCmp1WrDbgMQsbQi1c2VpPtXyIjs/49k\nv4VAl1TsAAByQrAD+q5cLi8sLKQDziYmJlZWVlo/eHZ2Nn3w7Oxs85iwtbW1GzdupC+1trbW\n+kfXR7mtrKy0+On117xx40bza7ZoUv3162PpGsfVpd/v7u6mr7znZdOfuLu72/H/1+7ubvqU\nhYWFelMPOtX7trCDHw1kWg2gC4deSVZXV5uvPMvLy/u+wr1795offO/evfqDi8XinntnZmYO\nbd6eNux5yszMTIvXbN2keuP3vbTWv09/xPb2dv1lt7e3G39QZ/9f9Wel57P1qd63hR38aCDL\nBDugK81JYk9uSG+ur6+nN9fX1/d9QPp9oVAIIWxubjY+eHJyMr25sbGRxo6dnZ1arbazs5MG\npo2NjdbNKxQK6Wtubm6mP2JPe1o8oHWT9o1xzf9r6bNWV1frd6UhLG15x/9f9ad0cKo7PqX8\n/+3cMU7rShgG0NkDEhFCYhk0SBRQhy4SbAB5AbTpKNNRBIUFhJoeJDpEBRUypRtEKnbALUbX\nshLbCQZx8+adUzmT8cxvF9GnydiwzgQ74FuWBrumU2o/xuOmVBHXlqpR5uPjI4QwGo3a56oO\nGKNMeUocc7FDuWrVXtKKwS4el3Hw8/Mzy7Lyq87Xled5U4fakhYr7DA1sM48FQt8y4oPWs5m\ns7e3t6IoHh8fz8/Pq6dUR7i+vj45OQkhjEajwWAQQtjZ2Zmbq1ZTAbXlVRuXdlilpKahqi1X\nV1enp6fv7++bm5uz2azX602n0+Pj4x+8rmjFW11t+dLUwFr79SgJJGWVX5LFTWyheRnp5uYm\n/vsZ9fv9cmtah5+y2m+rjUs7rFJS01DVljzPw99/Y6fTaaist/3UdX1+8VZ3mxpYZ1bsgG9Z\numIXV6qyLBsMBhsbG1tbW71eL7QuI4UQnp+fX15e7u/vLy8vsywbj8erzLVieV9asVu9pPYV\nuxDC0dHR9vb2eDwuD1qm63BdHW61N9tBan49SgJJWfpLMtchPg0aWpeRmk5f3A+3YnnVvWhx\nC91kMmkac24TXntJTce1LXGhLj5mW32QovN1tTeucqs7TA2sM++xA37D6+trCKEoiouLi5Zu\n8W1qDw8P8WM8iM8ZhBD29vZCCMPhsCiK2HJ3d1d9i1uTs7OzeEpRFMPhMISwu7tbO2bZIbYv\nLWnRbDZr+urg4CCEcHh4GELY398v2ztfV62lt7pa4c9ODfx7/zpZAv9tS39J4jLVorkdZvG4\nfENHbc/Puj1k1R1vTeXNnTW3Gtf+Hrv2kqrFl2mv3+833ZzYZ/FFcd2u66u3erHCDlMD60yw\nA76lNmHMmUwmZWDK83zuv865EZ6ensqoEfvPjXZ7e1sGlMlk0h5BysHjf45ZllVfd1wdMz4e\nUduhpaRq8Xmex8Jagl2MibV/fXa7rjntt3qxwg5TA+vMwxNAyjwcAPyv2GMHAJAIwQ4AIBGC\nHQBAIuyxAwBIhBU7AIBECHYAAIkQ7AAAEiHYAQAkQrADAEiEYAcAkAjBDgAgEYIdAEAiBDsA\ngEQIdgAAiRDsAAASIdgBACRCsAMASIRgBwCQCMEOACARgh0AQCIEOwCARAh2AACJEOwAABIh\n2AEAJEKwAwBIhGAHAJAIwQ4AIBGCHQBAIgQ7AIBECHYAAIn4AxXYlIlsR9wvAAAAAElFTkSu\nQmCC",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC and AUC for Undersampled NB\n",
    "probs <- predict(NBModelUnder, tst, type=\"raw\")\n",
    "library(ROCR)\n",
    "NBPred <- prediction(probs[,2], caravan_df_testNB$Number_of_mobile_home_policies)\n",
    "NBPerf <- performance(NBPred, \"tpr\", \"fpr\")\n",
    "plot(NBPerf, colorize=TRUE)\n",
    "abline(a=0, b=1, lty=2, lwd=3, col=\"black\")\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$NBU[i] <- probs[i]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 282,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "An object of class \"performance\"\n",
       "Slot \"x.name\":\n",
       "[1] \"None\"\n",
       "\n",
       "Slot \"y.name\":\n",
       "[1] \"Area under the ROC curve\"\n",
       "\n",
       "Slot \"alpha.name\":\n",
       "[1] \"none\"\n",
       "\n",
       "Slot \"x.values\":\n",
       "list()\n",
       "\n",
       "Slot \"y.values\":\n",
       "[[1]]\n",
       "[1] 0.6755224\n",
       "\n",
       "\n",
       "Slot \"alpha.values\":\n",
       "list()\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#AUC\n",
    "performance(NBPred, \"auc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 283,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<dl class=dl-horizontal>\n",
       "\t<dt>acc</dt>\n",
       "\t\t<dd>0.55175</dd>\n",
       "\t<dt>sens</dt>\n",
       "\t\t<dd>0.701680672268908</dd>\n",
       "\t<dt>spec</dt>\n",
       "\t\t<dd>0.542264752791069</dd>\n",
       "\t<dt>ppv</dt>\n",
       "\t\t<dd>0.0884065643197459</dd>\n",
       "\t<dt>npv</dt>\n",
       "\t\t<dd>0.966366650876362</dd>\n",
       "\t<dt>lor</dt>\n",
       "\t\t<dd>1.02477733722603</dd>\n",
       "</dl>\n"
      ],
      "text/latex": [
       "\\begin{description*}\n",
       "\\item[acc] 0.55175\n",
       "\\item[sens] 0.701680672268908\n",
       "\\item[spec] 0.542264752791069\n",
       "\\item[ppv] 0.0884065643197459\n",
       "\\item[npv] 0.966366650876362\n",
       "\\item[lor] 1.02477733722603\n",
       "\\end{description*}\n"
      ],
      "text/markdown": [
       "acc\n",
       ":   0.55175sens\n",
       ":   0.701680672268908spec\n",
       ":   0.542264752791069ppv\n",
       ":   0.0884065643197459npv\n",
       ":   0.966366650876362lor\n",
       ":   1.02477733722603\n",
       "\n"
      ],
      "text/plain": [
       "       acc       sens       spec        ppv        npv        lor \n",
       "0.55175000 0.70168067 0.54226475 0.08840656 0.96636665 1.02477734 \n",
       "attr(,\"negative\")\n",
       "[1] \"Not Purchased\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Corresponding Performance Measures\n",
    "NBPrediction <- factor(as.factor(prediction), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "NBActual <- factor(as.factor(caravan_df_testNB$Number_of_mobile_home_policies), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "library(crossval)\n",
    "CMNB <- confusionMatrix(NBActual, NBPrediction, negative = \"Not Purchased\" )\n",
    "diagnosticErrors(CMNB)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Naive Bayes Using Simplest Approach of Over Sampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 284,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "'Ratio of 1s to 0s- 1:15.7298850574713'"
      ],
      "text/latex": [
       "'Ratio of 1s to 0s- 1:15.7298850574713'"
      ],
      "text/markdown": [
       "'Ratio of 1s to 0s- 1:15.7298850574713'"
      ],
      "text/plain": [
       "[1] \"Ratio of 1s to 0s- 1:15.7298850574713\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#First let's recall the ratio of 1s and 0s in our original training dataset\n",
    "paste0(\"Ratio of 1s to 0s- 1:\", as.numeric(table(caravan_df_trainNB$Number_of_mobile_home_policies))[1]/as.numeric(table(caravan_df_trainNB$Number_of_mobile_home_policies))[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 285,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead><tr><th scope=col>Number_of_mobile_home_policies</th><th scope=col>Count</th></tr></thead>\n",
       "<tbody>\n",
       "\t<tr><td>0   </td><td>5474</td></tr>\n",
       "\t<tr><td>1   </td><td>2436</td></tr>\n",
       "</tbody>\n",
       "</table>\n"
      ],
      "text/latex": [
       "\\begin{tabular}{r|ll}\n",
       " Number\\_of\\_mobile\\_home\\_policies & Count\\\\\n",
       "\\hline\n",
       "\t 0    & 5474\\\\\n",
       "\t 1    & 2436\\\\\n",
       "\\end{tabular}\n"
      ],
      "text/markdown": [
       "\n",
       "Number_of_mobile_home_policies | Count | \n",
       "|---|---|\n",
       "| 0    | 5474 | \n",
       "| 1    | 2436 | \n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "  Number_of_mobile_home_policies Count\n",
       "1 0                              5474 \n",
       "2 1                              2436 "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Now let's duplicate the observations with dependent variable = 1 to make the ratio approximately 1:2\n",
    "caravan_df_train_NB_OverDummy <- caravan_df_trainNB[caravan_df_trainNB$Number_of_mobile_home_policies==1,]\n",
    "caravan_df_train_NB_Over <- NULL\n",
    "for (i in 1:7){\n",
    "    caravan_df_train_NB_Over <- rbind(caravan_df_train_NB_Over, caravan_df_train_NB_OverDummy)\n",
    "}\n",
    "caravan_df_train_NB_Over <- rbind(caravan_df_train_NB_Over, caravan_df_trainNB[caravan_df_trainNB$Number_of_mobile_home_policies==0,])\n",
    "\n",
    "#Let's verify the number of 1s and 0s in our oversampled data\n",
    "oversampled.Frequency.Number_of_mobile_home_policies <- data.frame(Number_of_mobile_home_policies = levels(as.factor(caravan_df_train_NB_Over$Number_of_mobile_home_policies)), Count = as.numeric(table(caravan_df_train_NB_Over$Number_of_mobile_home_policies)))\n",
    "oversampled.Frequency.Number_of_mobile_home_policies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 286,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "'Ratio of 1s to 0s After Oversampling- 1:2.24712643678161'"
      ],
      "text/latex": [
       "'Ratio of 1s to 0s After Oversampling- 1:2.24712643678161'"
      ],
      "text/markdown": [
       "'Ratio of 1s to 0s After Oversampling- 1:2.24712643678161'"
      ],
      "text/plain": [
       "[1] \"Ratio of 1s to 0s After Oversampling- 1:2.24712643678161\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Let's verify the ratio of 1s to 0s after over sampling\n",
    "paste0(\"Ratio of 1s to 0s After Oversampling- 1:\", as.numeric(table(caravan_df_train_NB_Over$Number_of_mobile_home_policies))[1]/as.numeric(table(caravan_df_train_NB_Over$Number_of_mobile_home_policies))[2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\n",
       "Naive Bayes Classifier for Discrete Predictors\n",
       "\n",
       "Call:\n",
       "naiveBayes.default(x = X, y = Y, laplace = laplace)\n",
       "\n",
       "A-priori probabilities:\n",
       "Y\n",
       "        0         1 \n",
       "0.6920354 0.3079646 \n",
       "\n",
       "Conditional probabilities:\n",
       "   car_policies\n",
       "Y       [,1]     [,2]\n",
       "  0 2.859335 2.914510\n",
       "  1 4.718391 2.418343\n",
       "\n",
       "   Income_._30\n",
       "Y       [,1]     [,2]\n",
       "  0 2.615638 2.097796\n",
       "  1 1.913793 1.766433\n",
       "\n",
       "   moped_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.22415053 0.8288564\n",
       "  1 0.07183908 0.4831570\n",
       "\n",
       "   fire_policies\n",
       "Y       [,1]     [,2]\n",
       "  0 1.782974 1.875056\n",
       "  1 2.531609 1.804462\n",
       "\n",
       "   Lower_level_education\n",
       "Y       [,1]     [,2]\n",
       "  0 4.624954 2.288806\n",
       "  1 3.747126 2.287205\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Running NB on Oversampled Data\n",
    "library(e1071)\n",
    "# Can handle both categorical and numeric input,\n",
    "# but output must be categorical\n",
    "NBModelOver <- naiveBayes(as.factor(Number_of_mobile_home_policies)~car_policies+Income_._30+moped_policies+fire_policies+Lower_level_education, data=caravan_df_train_NB_Over)\n",
    "NBModelOver"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "      predicted\n",
       "actual    0    1\n",
       "     0 2736 1026\n",
       "     1  107  131"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#NB Prediction for oversampling\n",
    "tst <- caravan_df_testNB[,-86]\n",
    "prediction <- predict(NBModelOver, newdata = tst)\n",
    "table(caravan_df_testNB$Number_of_mobile_home_policies, prediction, dnn=list('actual','predicted'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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\nBPEboTy5Wt9qtRMAAAAASUVORK5CYII=",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC and AUC for Oversampled NB\n",
    "probs <- predict(NBModelOver, tst, type=\"raw\")\n",
    "library(ROCR)\n",
    "NBPred <- prediction(probs[,2], caravan_df_testNB$Number_of_mobile_home_policies)\n",
    "NBPerf <- performance(NBPred, \"tpr\", \"fpr\")\n",
    "plot(NBPerf, colorize=TRUE)\n",
    "abline(a=0, b=1, lty=2, lwd=3, col=\"black\")\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$NBO[i] <- probs[i]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 290,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "An object of class \"performance\"\n",
       "Slot \"x.name\":\n",
       "[1] \"None\"\n",
       "\n",
       "Slot \"y.name\":\n",
       "[1] \"Area under the ROC curve\"\n",
       "\n",
       "Slot \"alpha.name\":\n",
       "[1] \"none\"\n",
       "\n",
       "Slot \"x.values\":\n",
       "list()\n",
       "\n",
       "Slot \"y.values\":\n",
       "[[1]]\n",
       "[1] 0.6766459\n",
       "\n",
       "\n",
       "Slot \"alpha.values\":\n",
       "list()\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#AUC\n",
    "performance(NBPred, \"auc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 291,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<dl class=dl-horizontal>\n",
       "\t<dt>acc</dt>\n",
       "\t\t<dd>0.71675</dd>\n",
       "\t<dt>sens</dt>\n",
       "\t\t<dd>0.550420168067227</dd>\n",
       "\t<dt>spec</dt>\n",
       "\t\t<dd>0.727272727272727</dd>\n",
       "\t<dt>ppv</dt>\n",
       "\t\t<dd>0.113223854796888</dd>\n",
       "\t<dt>npv</dt>\n",
       "\t\t<dd>0.962363700316567</dd>\n",
       "\t<dt>lor</dt>\n",
       "\t\t<dd>1.18319774175097</dd>\n",
       "</dl>\n"
      ],
      "text/latex": [
       "\\begin{description*}\n",
       "\\item[acc] 0.71675\n",
       "\\item[sens] 0.550420168067227\n",
       "\\item[spec] 0.727272727272727\n",
       "\\item[ppv] 0.113223854796888\n",
       "\\item[npv] 0.962363700316567\n",
       "\\item[lor] 1.18319774175097\n",
       "\\end{description*}\n"
      ],
      "text/markdown": [
       "acc\n",
       ":   0.71675sens\n",
       ":   0.550420168067227spec\n",
       ":   0.727272727272727ppv\n",
       ":   0.113223854796888npv\n",
       ":   0.962363700316567lor\n",
       ":   1.18319774175097\n",
       "\n"
      ],
      "text/plain": [
       "      acc      sens      spec       ppv       npv       lor \n",
       "0.7167500 0.5504202 0.7272727 0.1132239 0.9623637 1.1831977 \n",
       "attr(,\"negative\")\n",
       "[1] \"Not Purchased\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Corresponding Performance Measures\n",
    "NBPrediction <- factor(as.factor(prediction), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "NBActual <- factor(as.factor(caravan_df_testNB$Number_of_mobile_home_policies), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "library(crossval)\n",
    "CMNB <- confusionMatrix(NBActual, NBPrediction, negative = \"Not Purchased\" )\n",
    "diagnosticErrors(CMNB)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Naive Bayes Using SMOTE method for oversampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 292,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message:\n",
      "\"package 'DMwR' was built under R version 3.3.3\"Loading required package: lattice\n"
     ]
    }
   ],
   "source": [
    "library(DMwR)\n",
    "set.seed(123457)\n",
    "caravan_df_trainNB$Number_of_mobile_home_policies <- as.factor(caravan_df_trainNB$Number_of_mobile_home_policies)\n",
    "caravan_df_train_NB_SMOTE <- SMOTE(Number_of_mobile_home_policies~., caravan_df_trainNB, perc.over = 600, perc.under = 262.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 293,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<thead><tr><th scope=col>Number_of_mobile_home_policies</th><th scope=col>Count</th></tr></thead>\n",
       "<tbody>\n",
       "\t<tr><td>0   </td><td>5474</td></tr>\n",
       "\t<tr><td>1   </td><td>2436</td></tr>\n",
       "</tbody>\n",
       "</table>\n"
      ],
      "text/latex": [
       "\\begin{tabular}{r|ll}\n",
       " Number\\_of\\_mobile\\_home\\_policies & Count\\\\\n",
       "\\hline\n",
       "\t 0    & 5474\\\\\n",
       "\t 1    & 2436\\\\\n",
       "\\end{tabular}\n"
      ],
      "text/markdown": [
       "\n",
       "Number_of_mobile_home_policies | Count | \n",
       "|---|---|\n",
       "| 0    | 5474 | \n",
       "| 1    | 2436 | \n",
       "\n",
       "\n"
      ],
      "text/plain": [
       "  Number_of_mobile_home_policies Count\n",
       "1 0                              5474 \n",
       "2 1                              2436 "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Let's verify the number of 1s and 0s in our SMOTE data\n",
    "SMOTE.Frequency.Number_of_mobile_home_policies <- data.frame(Number_of_mobile_home_policies = levels(as.factor(caravan_df_train_NB_SMOTE$Number_of_mobile_home_policies)), Count = as.numeric(table(caravan_df_train_NB_SMOTE$Number_of_mobile_home_policies)))\n",
    "SMOTE.Frequency.Number_of_mobile_home_policies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 294,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\n",
       "Naive Bayes Classifier for Discrete Predictors\n",
       "\n",
       "Call:\n",
       "naiveBayes.default(x = X, y = Y, laplace = laplace)\n",
       "\n",
       "A-priori probabilities:\n",
       "Y\n",
       "        0         1 \n",
       "0.6920354 0.3079646 \n",
       "\n",
       "Conditional probabilities:\n",
       "   Customer_Subtype\n",
       "Y              1            2            3            4            5\n",
       "  0 0.0200949945 0.0138838144 0.0358056266 0.0087687249 0.0076726343\n",
       "  1 0.0217569787 0.0102627258 0.1243842365 0.0032840722 0.0036945813\n",
       "   Customer_Subtype\n",
       "Y              6            7            8            9           10\n",
       "  0 0.0219218122 0.0084033613 0.0575447570 0.0476799415 0.0325173548\n",
       "  1 0.0188834154 0.0049261084 0.0845648604 0.1096059113 0.0151888342\n",
       "   Customer_Subtype\n",
       "Y             11           12           13           15           16\n",
       "  0 0.0235659481 0.0157106321 0.0334307636 0.0007307271 0.0032882718\n",
       "  1 0.0164203612 0.1100164204 0.0229885057 0.0000000000 0.0000000000\n",
       "   Customer_Subtype\n",
       "Y             17           18           19           20           21\n",
       "  0 0.0014614541 0.0025575448 0.0007307271 0.0040189989 0.0023748630\n",
       "  1 0.0000000000 0.0000000000 0.0000000000 0.0032840722 0.0000000000\n",
       "   Customer_Subtype\n",
       "Y             22           23           24           25           26\n",
       "  0 0.0168067227 0.0409207161 0.0341614907 0.0111435879 0.0084033613\n",
       "  1 0.0073891626 0.0065681445 0.0090311987 0.0024630542 0.0020525452\n",
       "   Customer_Subtype\n",
       "Y             27           28           29           30           31\n",
       "  0 0.0074899525 0.0052977713 0.0188162221 0.0219218122 0.0341614907\n",
       "  1 0.0020525452 0.0000000000 0.0041050903 0.0073891626 0.0082101806\n",
       "   Customer_Subtype\n",
       "Y             32           33           34           35           36\n",
       "  0 0.0230179028 0.1359152357 0.0363536719 0.0359883084 0.0394592620\n",
       "  1 0.0114942529 0.0693760263 0.0147783251 0.0123152709 0.1018062397\n",
       "   Customer_Subtype\n",
       "Y             37           38           39           40           41\n",
       "  0 0.0215564487 0.0595542565 0.0580928023 0.0133357691 0.0354402631\n",
       "  1 0.0184729064 0.1338259442 0.0316091954 0.0000000000 0.0077996716\n",
       "\n",
       "   Number_of_houses\n",
       "Y       [,1]      [,2]\n",
       "  0 1.115638 0.4170234\n",
       "  1 1.053683 0.1939946\n",
       "\n",
       "   Avg_size_household\n",
       "Y       [,1]      [,2]\n",
       "  0 2.695835 0.7840644\n",
       "  1 2.629326 0.5736333\n",
       "\n",
       "   Avg_age\n",
       "Y       [,1]      [,2]\n",
       "  0 2.969858 0.8239990\n",
       "  1 3.081080 0.6442085\n",
       "\n",
       "   Customer_main_type\n",
       "Y             1           2           3           4           5           6\n",
       "  0 0.086225795 0.087869931 0.152904640 0.008768725 0.098282791 0.032334673\n",
       "  1 0.161740558 0.111247947 0.278325123 0.000000000 0.025041051 0.006978654\n",
       "   Customer_main_type\n",
       "Y             7           8           9          10\n",
       "  0 0.097917428 0.269272927 0.117647059 0.048776032\n",
       "  1 0.034482759 0.217569787 0.156403941 0.008210181\n",
       "\n",
       "   Roman_catholic\n",
       "Y        [,1]      [,2]\n",
       "  0 0.6782974 0.9764710\n",
       "  1 0.5601797 0.7266698\n",
       "\n",
       "   Protestant\n",
       "Y       [,1]     [,2]\n",
       "  0 4.615638 1.745936\n",
       "  1 4.941674 1.181611\n",
       "\n",
       "   Other_religion\n",
       "Y       [,1]     [,2]\n",
       "  0 1.062843 1.015176\n",
       "  1 1.292242 1.006591\n",
       "\n",
       "   No_religion\n",
       "Y       [,1]     [,2]\n",
       "  0 3.286262 1.583559\n",
       "  1 2.739549 1.122596\n",
       "\n",
       "   Married\n",
       "Y       [,1]     [,2]\n",
       "  0 6.188162 1.916787\n",
       "  1 6.753370 1.086210\n",
       "\n",
       "   Living_together\n",
       "Y        [,1]      [,2]\n",
       "  0 0.8913043 1.0044962\n",
       "  1 0.7804403 0.6624655\n",
       "\n",
       "   Other_relation\n",
       "Y       [,1]     [,2]\n",
       "  0 2.272561 1.728690\n",
       "  1 1.748640 1.116855\n",
       "\n",
       "   Singles\n",
       "Y       [,1]     [,2]\n",
       "  0 1.846730 1.779474\n",
       "  1 2.087455 1.331776\n",
       "\n",
       "   Household_without_children\n",
       "Y       [,1]     [,2]\n",
       "  0 3.197662 1.615753\n",
       "  1 2.947642 1.287264\n",
       "\n",
       "   Household_with_children\n",
       "Y       [,1]     [,2]\n",
       "  0 4.369931 2.002278\n",
       "  1 4.428741 1.457922\n",
       "\n",
       "   High_level_education\n",
       "Y       [,1]     [,2]\n",
       "  0 1.395689 1.601943\n",
       "  1 2.532590 1.798877\n",
       "\n",
       "   Medium_level_education\n",
       "Y       [,1]     [,2]\n",
       "  0 3.349470 1.757678\n",
       "  1 3.626454 1.265367\n",
       "\n",
       "   Lower_level_education\n",
       "Y       [,1]     [,2]\n",
       "  0 4.622031 2.288884\n",
       "  1 3.328702 1.996052\n",
       "\n",
       "   High_status\n",
       "Y       [,1]     [,2]\n",
       "  0 1.834490 1.783763\n",
       "  1 2.652913 1.806395\n",
       "\n",
       "   Entrepreneur\n",
       "Y        [,1]      [,2]\n",
       "  0 0.3854585 0.7685441\n",
       "  1 0.3567744 0.5785618\n",
       "\n",
       "   Farmer\n",
       "Y        [,1]      [,2]\n",
       "  0 0.5199123 1.0666043\n",
       "  1 0.2611123 0.5295373\n",
       "\n",
       "   Middle_management\n",
       "Y       [,1]     [,2]\n",
       "  0 2.881805 1.828801\n",
       "  1 3.315406 1.511860\n",
       "\n",
       "   Skilled_labourers\n",
       "Y       [,1]     [,2]\n",
       "  0 2.244611 1.748663\n",
       "  1 1.704646 1.261232\n",
       "\n",
       "   Unskilled_labourers\n",
       "Y       [,1]     [,2]\n",
       "  0 2.348739 1.708378\n",
       "  1 1.853630 1.312642\n",
       "\n",
       "   Social_class_A\n",
       "Y       [,1]     [,2]\n",
       "  0 1.556997 1.691798\n",
       "  1 2.028777 1.816279\n",
       "\n",
       "   Social_class_B1\n",
       "Y       [,1]     [,2]\n",
       "  0 1.602850 1.329684\n",
       "  1 2.328288 1.752755\n",
       "\n",
       "   Social_class_B2\n",
       "Y       [,1]     [,2]\n",
       "  0 2.220497 1.549173\n",
       "  1 1.838764 1.268231\n",
       "\n",
       "   Social_class_C\n",
       "Y       [,1]     [,2]\n",
       "  0 3.778590 1.920760\n",
       "  1 3.066755 1.783432\n",
       "\n",
       "   Social_class_D\n",
       "Y        [,1]      [,2]\n",
       "  0 1.0842163 1.3193092\n",
       "  1 0.5969544 0.8117237\n",
       "\n",
       "   Rented_house\n",
       "Y       [,1]     [,2]\n",
       "  0 4.230179 3.089563\n",
       "  1 3.274806 2.725288\n",
       "\n",
       "   Home_owners\n",
       "Y       [,1]     [,2]\n",
       "  0 4.777311 3.089846\n",
       "  1 5.725194 2.725288\n",
       "\n",
       "   X1_car\n",
       "Y       [,1]     [,2]\n",
       "  0 6.035258 1.556430\n",
       "  1 6.462848 1.316234\n",
       "\n",
       "   X2_cars\n",
       "Y       [,1]     [,2]\n",
       "  0 1.337961 1.190802\n",
       "  1 1.610158 1.118974\n",
       "\n",
       "   No_car\n",
       "Y       [,1]     [,2]\n",
       "  0 1.938802 1.595060\n",
       "  1 1.266319 1.031736\n",
       "\n",
       "   National_Health_Service\n",
       "Y       [,1]     [,2]\n",
       "  0 6.317136 1.963371\n",
       "  1 5.383469 1.580265\n",
       "\n",
       "   Private_health_insurance\n",
       "Y       [,1]     [,2]\n",
       "  0 2.688528 1.966741\n",
       "  1 3.616531 1.580265\n",
       "\n",
       "   Income_._30\n",
       "Y       [,1]     [,2]\n",
       "  0 2.597917 2.123280\n",
       "  1 1.777300 1.403791\n",
       "\n",
       "   Income_30.45.000\n",
       "Y       [,1]     [,2]\n",
       "  0 3.547315 1.921169\n",
       "  1 3.818792 1.328543\n",
       "\n",
       "   Income_45.75.000\n",
       "Y       [,1]     [,2]\n",
       "  0 2.704969 1.958949\n",
       "  1 3.354447 1.489586\n",
       "\n",
       "   Income_75.122.000\n",
       "Y        [,1]      [,2]\n",
       "  0 0.7723785 1.1639245\n",
       "  1 0.6906085 0.9041639\n",
       "\n",
       "   Income_.123.000\n",
       "Y        [,1]      [,2]\n",
       "  0 0.1830471 0.5162421\n",
       "  1 0.2029904 0.3774841\n",
       "\n",
       "   Average_income\n",
       "Y       [,1]      [,2]\n",
       "  0 3.749178 1.3139278\n",
       "  1 4.146918 0.9040489\n",
       "\n",
       "   Purchasing_power\n",
       "Y       [,1]     [,2]\n",
       "  0 4.243515 2.009042\n",
       "  1 4.934809 1.521282\n",
       "\n",
       "   private_third_party_insurance\n",
       "Y        [,1]      [,2]\n",
       "  0 0.7643405 0.9554291\n",
       "  1 1.1621997 0.8347523\n",
       "\n",
       "   third_party_insurance_firms\n",
       "Y         [,1]      [,2]\n",
       "  0 0.04384362 0.3841072\n",
       "  1 0.02197903 0.2046290\n",
       "\n",
       "   third_party_insurane_agriculture\n",
       "Y         [,1]      [,2]\n",
       "  0 0.07526489 0.5131188\n",
       "  1 0.01993693 0.2330839\n",
       "\n",
       "   car_policies\n",
       "Y       [,1]     [,2]\n",
       "  0 2.916880 2.905264\n",
       "  1 3.736696 2.354420\n",
       "\n",
       "   delivery_van_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.05224699 0.5515716\n",
       "  1 0.01656358 0.2679266\n",
       "\n",
       "   Contribution_motorcycle.scooter\n",
       "Y        [,1]      [,2]\n",
       "  0 0.1739130 0.8974897\n",
       "  1 0.5735371 1.2851443\n",
       "\n",
       "   lorry_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01041286 0.2594483\n",
       "  1 0.00000000 0.0000000\n",
       "\n",
       "   trailer_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01753745 0.1960306\n",
       "  1 0.01695570 0.1489186\n",
       "\n",
       "   tractor_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.10650347 0.6517638\n",
       "  1 0.03677815 0.3674557\n",
       "\n",
       "   agricultural_machines_policies\n",
       "Y       [,1]      [,2]\n",
       "  0 0.016076 0.2530953\n",
       "  1 0.000000 0.0000000\n",
       "\n",
       "   moped_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.22981366 0.8422966\n",
       "  1 0.04043929 0.3120739\n",
       "\n",
       "   life_insurances\n",
       "Y        [,1]      [,2]\n",
       "  0 0.1972963 0.9106753\n",
       "  1 0.1616492 0.7092027\n",
       "\n",
       "   private_accident_insurance_policies\n",
       "Y         [,1]       [,2]\n",
       "  0 0.01607600 0.25019098\n",
       "  1 0.00344107 0.07389827\n",
       "\n",
       "   family_accidents_insurance_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01479722 0.1852408\n",
       "  1 0.02628343 0.2322864\n",
       "\n",
       "   disability_insurance_policies\n",
       "Y         [,1]     [,2]\n",
       "  0 0.02082572 0.354972\n",
       "  1 0.55251523 1.389347\n",
       "\n",
       "   fire_policies\n",
       "Y       [,1]     [,2]\n",
       "  0 1.830106 1.904249\n",
       "  1 2.387554 1.540489\n",
       "\n",
       "   surfboard_policies\n",
       "Y           [,1]       [,2]\n",
       "  0 0.0001826818 0.01351598\n",
       "  1 0.0014094517 0.03317643\n",
       "\n",
       "   boat_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01223968 0.2050789\n",
       "  1 0.06092917 0.4068875\n",
       "\n",
       "   bicycle_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.02429668 0.1539827\n",
       "  1 0.02679309 0.1415423\n",
       "\n",
       "   property_insurance_policies\n",
       "Y          [,1]      [,2]\n",
       "  0 0.013335769 0.1853518\n",
       "  1 0.009936395 0.1028859\n",
       "\n",
       "   social_security_insurance_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.03872853 0.3631555\n",
       "  1 0.08200616 0.4473073\n",
       "\n",
       "   Number_of_private_third_party_insurance\n",
       "Y        [,1]      [,2]\n",
       "  0 0.3997077 0.4910006\n",
       "  1 0.5857137 0.4172350\n",
       "\n",
       "   Number_of_third_party_insurance_firms\n",
       "Y          [,1]       [,2]\n",
       "  0 0.015527950 0.13763695\n",
       "  1 0.008729663 0.08073886\n",
       "\n",
       "   Number_of_third_party_insurane_agriculture\n",
       "Y          [,1]       [,2]\n",
       "  0 0.021556449 0.14524332\n",
       "  1 0.005951269 0.06907893\n",
       "\n",
       "   Number_of_car_policies\n",
       "Y        [,1]      [,2]\n",
       "  0 0.5480453 0.5931986\n",
       "  1 0.6899238 0.4753705\n",
       "\n",
       "   Number_of_delivery_van_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.010595543 0.12343333\n",
       "  1 0.002760597 0.04465443\n",
       "\n",
       "   Number_of_motorcycle.scooter_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.04037267 0.2286279\n",
       "  1 0.12021256 0.2708585\n",
       "\n",
       "   Number_of_lorry_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.002922908 0.07640903\n",
       "  1 0.000000000 0.00000000\n",
       "\n",
       "   Number_of_trailer_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.010595543 0.11890961\n",
       "  1 0.009545984 0.08252344\n",
       "\n",
       "   Number_of_tractor_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.03854585 0.2546361\n",
       "  1 0.01001432 0.1029281\n",
       "\n",
       "   Number_of_agricultural_machines_policies\n",
       "Y          [,1]      [,2]\n",
       "  0 0.007489953 0.1356397\n",
       "  1 0.000000000 0.0000000\n",
       "\n",
       "   Number_of_moped_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.07544757 0.2769678\n",
       "  1 0.01290893 0.0964441\n",
       "\n",
       "   Number_of_life_insurances\n",
       "Y         [,1]      [,2]\n",
       "  0 0.07782243 0.3891568\n",
       "  1 0.07944526 0.3797082\n",
       "\n",
       "   Number_of_private_accident_insurance_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.005115090 0.07134322\n",
       "  1 0.001720535 0.03694914\n",
       "\n",
       "   Number_of_family_accidents_insurance_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.006576544 0.08083617\n",
       "  1 0.009834744 0.08595427\n",
       "\n",
       "   Number_of_disability_insurance_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.004201681 0.07514328\n",
       "  1 0.092085872 0.23155785\n",
       "\n",
       "   Number_of_fire_policies\n",
       "Y        [,1]      [,2]\n",
       "  0 0.5727073 0.5772283\n",
       "  1 0.6633997 0.4223776\n",
       "\n",
       "   Number_of_surfboard_policies\n",
       "Y           [,1]       [,2]\n",
       "  0 0.0001826818 0.01351598\n",
       "  1 0.0014094517 0.03317643\n",
       "\n",
       "   Number_of_boat_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.004018999 0.06609847\n",
       "  1 0.021597979 0.12943872\n",
       "\n",
       "   Number_of_bicycle_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.03014249 0.2032202\n",
       "  1 0.03797160 0.2207635\n",
       "\n",
       "   Number_of_property_insurance_policies\n",
       "Y          [,1]       [,2]\n",
       "  0 0.007489953 0.09036632\n",
       "  1 0.006757005 0.06673990\n",
       "\n",
       "   Number_of_social_security_insurance_policies\n",
       "Y         [,1]      [,2]\n",
       "  0 0.01169163 0.1075037\n",
       "  1 0.02574217 0.1361390\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Running NB on SMOTE Data\n",
    "library(e1071)\n",
    "# Can handle both categorical and numeric input,\n",
    "# but output must be categorical\n",
    "NBModelSMOTE <- naiveBayes(as.factor(Number_of_mobile_home_policies)~., data=caravan_df_train_NB_SMOTE)\n",
    "NBModelSMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 295,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "      predicted\n",
       "actual    0    1\n",
       "     0 1007 2755\n",
       "     1   34  204"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#NB Prediction for SMOTE\n",
    "tst <- caravan_df_testNB[,-86]\n",
    "prediction <- predict(NBModelSMOTE, newdata = tst)\n",
    "table(caravan_df_testNB$Number_of_mobile_home_policies, prediction, dnn=list('actual','predicted'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 296,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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\n7AAAEvH/A+ymxqL4mpzaAAAAAElFTkSuQmCC",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot ROC and AUC for SMOTE NB\n",
    "probs <- predict(NBModelSMOTE, tst, type=\"raw\")\n",
    "library(ROCR)\n",
    "NBPred <- prediction(probs[,2], caravan_df_testNB$Number_of_mobile_home_policies)\n",
    "NBPerf <- performance(NBPred, \"tpr\", \"fpr\")\n",
    "plot(NBPerf, colorize=TRUE)\n",
    "abline(a=0, b=1, lty=2, lwd=3, col=\"black\")\n",
    "for (i in 1:4000){\n",
    "    probabilityDF$NBS[i] <- probs[i]\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 297,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "An object of class \"performance\"\n",
       "Slot \"x.name\":\n",
       "[1] \"None\"\n",
       "\n",
       "Slot \"y.name\":\n",
       "[1] \"Area under the ROC curve\"\n",
       "\n",
       "Slot \"alpha.name\":\n",
       "[1] \"none\"\n",
       "\n",
       "Slot \"x.values\":\n",
       "list()\n",
       "\n",
       "Slot \"y.values\":\n",
       "[[1]]\n",
       "[1] 0.6028541\n",
       "\n",
       "\n",
       "Slot \"alpha.values\":\n",
       "list()\n"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#AUC\n",
    "performance(NBPred, \"auc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<dl class=dl-horizontal>\n",
       "\t<dt>acc</dt>\n",
       "\t\t<dd>0.30275</dd>\n",
       "\t<dt>sens</dt>\n",
       "\t\t<dd>0.857142857142857</dd>\n",
       "\t<dt>spec</dt>\n",
       "\t\t<dd>0.267676767676768</dd>\n",
       "\t<dt>ppv</dt>\n",
       "\t\t<dd>0.0689422102061507</dd>\n",
       "\t<dt>npv</dt>\n",
       "\t\t<dd>0.967339097022094</dd>\n",
       "\t<dt>lor</dt>\n",
       "\t\t<dd>0.785317640359603</dd>\n",
       "</dl>\n"
      ],
      "text/latex": [
       "\\begin{description*}\n",
       "\\item[acc] 0.30275\n",
       "\\item[sens] 0.857142857142857\n",
       "\\item[spec] 0.267676767676768\n",
       "\\item[ppv] 0.0689422102061507\n",
       "\\item[npv] 0.967339097022094\n",
       "\\item[lor] 0.785317640359603\n",
       "\\end{description*}\n"
      ],
      "text/markdown": [
       "acc\n",
       ":   0.30275sens\n",
       ":   0.857142857142857spec\n",
       ":   0.267676767676768ppv\n",
       ":   0.0689422102061507npv\n",
       ":   0.967339097022094lor\n",
       ":   0.785317640359603\n",
       "\n"
      ],
      "text/plain": [
       "       acc       sens       spec        ppv        npv        lor \n",
       "0.30275000 0.85714286 0.26767677 0.06894221 0.96733910 0.78531764 \n",
       "attr(,\"negative\")\n",
       "[1] \"Not Purchased\""
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Corresponding Performance Measures\n",
    "NBPrediction <- factor(as.factor(prediction), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "NBActual <- factor(as.factor(caravan_df_testNB$Number_of_mobile_home_policies), c(0, 1), labels = c(\"Not Purchased\", \"Purchased\"))\n",
    "library(crossval)\n",
    "CMNB <- confusionMatrix(NBActual, NBPrediction, negative = \"Not Purchased\" )\n",
    "diagnosticErrors(CMNB)\n",
    "write.csv(probabilityDF, file = \"probabilityDF.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### 6. Classification using Bagging - Unbalanced Data, Undersampled Data, Oversampled Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Bagging using unbalanced data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\n",
       "Call:\n",
       " randomForest(formula = as.factor(Number_of_mobile_home_policies) ~      ., data = TrainDataset, mtry = 85, importance = TRUE) \n",
       "               Type of random forest: classification\n",
       "                     Number of trees: 500\n",
       "No. of variables tried at each split: 85\n",
       "\n",
       "        OOB estimate of  error rate: 7.75%\n",
       "Confusion matrix:\n",
       "     0   1 class.error\n",
       "0 5346 128  0.02338327\n",
       "1  323  25  0.92816092"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Bagging using unbalanced data\n",
    "library(randomForest)\n",
    "library(plyr)\n",
    "library(caret)\n",
    "TrainDataset <- read.csv('C:\\\\Users\\\\Jayant\\\\Desktop\\\\DMPA\\\\Project\\\\Data_sets\\\\Data_sets\\\\Train.csv')\n",
    "\n",
    "TestDataset <- read.csv('C:\\\\Users\\\\Jayant\\\\Desktop\\\\DMPA\\\\Project\\\\Data_sets\\\\Data_sets\\\\Test_with_Targer_Variable.csv')\n",
    "set.seed(123457)\n",
    "\n",
    "#Bagging\n",
    "bag.train <- randomForest(as.factor(Number_of_mobile_home_policies) ~ ., data = TrainDataset, mtry = 85, importance = TRUE)\n",
    "bag.train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#Bagging Prediction For unbalanced data\n",
    "yhat.bag <- predict(bag.train, newdata = TestDataset)\n",
    "PredictedClass <- ifelse(yhat.bag > 0.5, 1, 0)\n",
    "\n",
    "#Confusion Matrix\n",
    "actual <- (TestDataset$Number_of_mobile_home_policies)\n",
    "actual <- as.factor(actual)\n",
    "PredictedClass <- as.factor(PredictedClass)\n",
    "levels(PredictedClass)\n",
    "str(PredictedClass)\n",
    "str(actual)\n",
    "\n",
    "confusionMatrix(PredictedClass, actual, positive = \"1\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Bagging using undersampled data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "data_balanced_under <- ovun.sample(Number_of_mobile_home_policies ~., data = TrainDataset, method = \"under\", N = 696, seed = 1)$data\n",
    "table(data_balanced_under$Number_of_mobile_home_policies)\n",
    "\n",
    "#Bagging on under sampled data\n",
    "bag.train.under <- randomForest(as.factor(Number_of_mobile_home_policies) ~ ., data = data_balanced_under, mtry = 86, importance = TRUE)\n",
    "bag.train.under\n",
    "\n",
    "#Prediction on under sampled data\n",
    "yhat.bag.under <- predict(bag.train.under, newdata = TestDataset)\n",
    "mean((yhat.bag.under-TestDataset)^2)\n",
    "\n",
    "PredictedClass.under <- ifelse(yhat.bag.under > 0.5, 1, 0)\n",
    "#Confusion Matrix of under sampled data\n",
    "actual <- (TestDataset$Number_of_mobile_home_policies)\n",
    "actual <- as.factor(actual)\n",
    "PredictedClass.under <- as.factor(PredictedClass.under)\n",
    "levels(PredictedClass.under)\n",
    "str(PredictedClass.under)\n",
    "str(actual)\n",
    "\n",
    "confusionMatrix(PredictedClass.under, actual, positive = \"1\")\n",
    "\n",
    "#Roc curve of under sampled data\n",
    "roc.curve(TestDataset$Number_of_mobile_home_policies, yhat.bag.under)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Bagging using over sampled data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "data_balanced_over <- ovun.sample(Number_of_mobile_home_policies ~. , data = TrainDataset, method = \"over\", N = 10948)$data\n",
    "#N refers to number of observations in the resulting balanced dataset. We had originally 5474 negative observations. \n",
    "\n",
    "#Bagging on Over sampled data\n",
    "bag.train.over <- randomForest(as.factor(Number_of_mobile_home_policies) ~ ., data = data_balanced_over, mtry = 86, importance = TRUE)\n",
    "bag.train.over\n",
    "\n",
    "#Prediction on over sampled data\n",
    "yhat.bag.over <- predict(bag.train.over, newdata = TestDataset)\n",
    "mean((yhat.bag.over-TestDataset)^2)\n",
    "\n",
    "PredictedClass.over <- ifelse(yhat.bag.over > 0.5, 1, 0)\n",
    "\n",
    "#Confusion Matrix of Over sampled data\n",
    "actual <- (TestDataset$Number_of_mobile_home_policies)\n",
    "actual <- as.factor(actual)\n",
    "PredictedClass.over <- as.factor(PredictedClass.over)\n",
    "levels(PredictedClass.over)\n",
    "str(PredictedClass.over)\n",
    "str(actual)\n",
    "\n",
    "confusionMatrix(PredictedClass.over, actual, positive = \"1\")\n",
    "\n",
    "#Roc curve of Over sampled data\n",
    "roc.curve(TestDataset$Number_of_mobile_home_policies, yhat.bag.over)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### 7. Classification using Boosting - Unbalanced Data, Undersampled Data, Oversampled Data, SMOTE - Including Business Cost For Each Model "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Boosting using unbalanced data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "library(tree)\n",
    "library(gbm)\n",
    "library(ROSE)\n",
    "library(caret)\n",
    "\n",
    "TrainDataset <- read.csv(\"/Users/sheenakhemani/Documents/SpringSem/Data Mining/Project/Data_sets/Train.csv\")\n",
    "attach(TrainDataset)\n",
    "\n",
    "TestDataset <- read.csv(\"/Users/sheenakhemani/Documents/SpringSem/Data Mining/Project/Data_sets/Test_with_Targer_Variable.csv\")\n",
    "attach(TestDataset)\n",
    "\n",
    "set.seed(123457)\n",
    "\n",
    "#Boosting on training set\n",
    "boost.train1 <- gbm(Number_of_mobile_home_policies ~., data = TrainDataset, distribution = \"bernoulli\", n.trees = 1000, interaction.depth = 4)\n",
    "summary(boost.train1)\n",
    "\n",
    "#car policies and fire policies seems to be the most important variables\n",
    "#producing partial dependance plots for the above 2 variables\n",
    "par(mfrow = c(1,2))\n",
    "plot(boost.train1, i = \"car_policies\")\n",
    "plot(boost.train1, i = \"fire_policies\")\n",
    "\n",
    "#Boosting to predict on test dataset\n",
    "yhat.boost.test1 <- predict.gbm(boost.train1, newdata = TestDataset, n.trees = 1000 )\n",
    "mean((yhat.boost.test1 - TestDataset)^2)\n",
    "#test MSE obtained is 34.138\n",
    "\n",
    "\n",
    "\n",
    "#Confusion Matrix\n",
    "PredictedClass.test1 <- ifelse(yhat.boost.test1 > 0.5, 1, 0)\n",
    "actual <- (TestDataset$Number_of_mobile_home_policies)\n",
    "actual <- as.factor(actual)\n",
    "PredictedClass.test1 <- as.factor(PredictedClass.test1)\n",
    "levels(PredictedClass.test1)\n",
    "str(PredictedClass.test1)\n",
    "str(actual)\n",
    "\n",
    "confusionMatrix(PredictedClass.test1, actual, positive = \"1\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Boosting on under sampled data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "data_balanced_under <- ovun.sample(Number_of_mobile_home_policies ~., data = TrainDataset, method = \"under\", N = 696, seed = 1)$data\n",
    "table(data_balanced_under$Number_of_mobile_home_policies)\n",
    "\n",
    "#Boosting on under sampled data\n",
    "boost.train.under <- gbm(Number_of_mobile_home_policies ~., data = data_balanced_under, distribution = \"bernoulli\", n.trees = 1000, interaction.depth = 4)\n",
    "summary(boost.train.under)\n",
    "\n",
    "#Predictions on under sampled data\n",
    "yhat.boost.test.under <- predict(boost.train.under, newdata = TestDataset, n.trees = 1000)\n",
    "mean((yhat.boost.test.under - TestDataset)^2)\n",
    "\n",
    "#Confusion Matrix - under sampled\n",
    "PredictedClass.under <- ifelse(yhat.boost.test.under > 0.5, 1, 0)\n",
    "\n",
    "actual <- (TestDataset$Number_of_mobile_home_policies)\n",
    "actual <- as.factor(actual)\n",
    "PredictedClass.under <- as.factor(PredictedClass.under)\n",
    "levels(PredictedClass.under)\n",
    "str(PredictedClass.under)\n",
    "str(actual)\n",
    "\n",
    "confusionMatrix(PredictedClass.under, actual, positive = \"1\")\n",
    "\n",
    "#Evaluating Accuracy of under sampled using ROC\n",
    "roc.curve(TestDataset$Number_of_mobile_home_policies, yhat.boost.test.under)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### Boosting using over sampled data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#Over Sampling \n",
    "data_balanced_over <- ovun.sample(Number_of_mobile_home_policies ~. , data = TrainDataset, method = \"over\", N = 10948)$data\n",
    "#N refers to number of observations in the resulting balanced dataset. We had originally 5474 negative observations. \n",
    "table(data_balanced_over$Number_of_mobile_home_policies)\n",
    "\n",
    "#Boosting on over sampled data\n",
    "boost.train.over <- gbm(Number_of_mobile_home_policies ~., data = data_balanced_over, distribution = \"bernoulli\", n.trees = 1000, interaction.depth = 4)\n",
    "summary(boost.train.over)\n",
    "\n",
    "#Predictions on over sampled data\n",
    "yhat.boost.test.over <- predict(boost.train.over, newdata = TestDataset, n.trees = 1000)\n",
    "mean((yhat.boost.test.over - TestDataset)^2)\n",
    "\n",
    "#Confusion Matrix - Over sampled\n",
    "PredictedClass.over <- ifelse(yhat.boost.test.over > 0.5, 1, 0)\n",
    "actual <- (TestDataset$Number_of_mobile_home_policies)\n",
    "actual <- as.factor(actual)\n",
    "PredictedClass.over <- as.factor(PredictedClass.over)\n",
    "levels(PredictedClass.over)\n",
    "str(PredictedClass.over)\n",
    "str(actual)\n",
    "\n",
    "confusionMatrix(PredictedClass.over, actual, positive = \"1\")\n",
    "\n",
    "#Evaluating Accuracy of over sampled using ROC\n",
    "roc.curve(TestDataset$Number_of_mobile_home_policies, yhat.boost.test.over)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### 8. Classification using Random Forest - Unbalanced Data, Undersampled Data, Oversampled Data, SMOTE - Including Business Cost For Each"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "###### Random Forest using unbalanced data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "library(randomForest)\n",
    "library(plyr)\n",
    "library(caret)\n",
    "TrainDataset <- read.csv(\"/Users/sheenakhemani/Documents/SpringSem/Data Mining/Project/Data_sets/Train.csv\")\n",
    "attach(TrainDataset)\n",
    "\n",
    "TestDataset <- read.csv(\"/Users/sheenakhemani/Documents/SpringSem/Data Mining/Project/Data_sets/Test_with_Targer_Variable.csv\")\n",
    "attach(TestDataset)\n",
    "set.seed(123457)\n",
    "\n",
    "#RandomForest\n",
    "rf.train <- randomForest(as.factor(Number_of_mobile_home_policies) ~., data = TrainDataset, mtry = 9, importance = TRUE)\n",
    "rf.train\n",
    "\n",
    "yhat.rf <- predict(rf.train, newdata = TestDataset)\n",
    "mean((yhat.rf-TestDataset)^2)\n",
    "# test MSE = 15.46\n",
    "\n",
    "importance(rf.train)\n",
    "\n",
    "\n",
    "#Confusion Matrix\n",
    "PredictedClass <- ifelse(yhat.rf > 0.5, 1, 0)\n",
    "actual <- (TestDataset$Number_of_mobile_home_policies)\n",
    "actual <- as.factor(actual)\n",
    "PredictedClass <- as.factor(PredictedClass)\n",
    "levels(PredictedClass)\n",
    "str(PredictedClass)\n",
    "str(actual)\n",
    "\n",
    "confusionMatrix(PredictedClass, actual, positive = \"1\")\n",
    "\n",
    "#ROC Curve\n",
    "roc.curve(TestDataset$Number_of_mobile_home_policies, yhat.rf, plotit = F)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "###### Random Forest using Under sampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#Under sampling \n",
    "data_balanced_under <- ovun.sample(Number_of_mobile_home_policies ~., data = TrainDataset, method = \"under\", N = 696, seed = 1)$data\n",
    "table(data_balanced_under$Number_of_mobile_home_policies)\n",
    "\n",
    "#RandomForest on under sampled data\n",
    "rf.train.under <- randomForest(as.factor(Number_of_mobile_home_policies) ~., data = data_balanced_under, mtry = 9, importance = TRUE)\n",
    "rf.train.under\n",
    "\n",
    "yhat.rf.under <- predict(rf.train.under, newdata = TestDataset)\n",
    "mean((yhat.rf.under-TestDataset)^2)\n",
    "# test MSE = 15.46\n",
    "\n",
    "importance(rf.train.under)\n",
    "\n",
    "#Confusion Matrix\n",
    "PredictedClass.under <- ifelse(yhat.rf.under > 0.5, 1, 0)\n",
    "actual <- (TestDataset$Number_of_mobile_home_policies)\n",
    "actual <- as.factor(actual)\n",
    "PredictedClass.under <- as.factor(PredictedClass.under)\n",
    "levels(PredictedClass.under)\n",
    "str(PredictedClass.under)\n",
    "str(actual)\n",
    "\n",
    "confusionMatrix(PredictedClass.under, actual, positive = \"1\")\n",
    "\n",
    "#ROC Curve\n",
    "roc.curve(TestDataset$Number_of_mobile_home_policies, yhat.rf.under, plotit = F)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "###### Random Forest Over Sampled Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#Over Sampling \n",
    "data_balanced_over <- ovun.sample(Number_of_mobile_home_policies ~. , data = TrainDataset, method = \"over\", N = 10948)$data\n",
    "\n",
    "#RandomForest on over sampled data\n",
    "rf.train.over <- randomForest(as.factor(Number_of_mobile_home_policies) ~., data = data_balanced_over, mtry = 9, importance = TRUE)\n",
    "rf.train.over\n",
    "\n",
    "yhat.rf.over <- predict(rf.train.over, newdata = TestDataset)\n",
    "mean((yhat.rf.over-TestDataset)^2)\n",
    "# test MSE = 15.46\n",
    "\n",
    "importance(rf.train.over)\n",
    "\n",
    "#Confusion Matrix\n",
    "PredictedClass.over <- ifelse(yhat.rf.over > 0.5, 1, 0)\n",
    "actual <- (TestDataset$Number_of_mobile_home_policies)\n",
    "actual <- as.factor(actual)\n",
    "PredictedClass.over <- as.factor(PredictedClass.over)\n",
    "levels(PredictedClass.over)\n",
    "str(PredictedClass.over)\n",
    "str(actual)\n",
    "\n",
    "confusionMatrix(PredictedClass.over, actual, positive = \"1\")\n",
    "\n",
    "#ROC Curve\n",
    "roc.curve(TestDataset$Number_of_mobile_home_policies, yhat.rf.over, plotit = F)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### 9. Summary of Performance Measures For All Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message:\n",
      "\"Ignoring unknown parameters: hjust, vjust\""
     ]
    },
    {
     "data": {},
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAIAAAByhViMAAAACXBIWXMAABJ0AAASdAHeZh94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\n/rXGOwAAAABJRU5ErkJggg==",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "library(ggrepel)\n",
    "library(ggplot2)\n",
    "library(plyr)\n",
    "Sensitivity <- c(0.012605042,\t0.647058824,\t0,\t1,\t0.991596639,\t0.995798319,\t0.067226891,\t0.298319328,\t0.445378151,\t0.142857143,\t0,\t0.617647059,\t0.621848739,\t0.075630252,\t0.117647059,\t0.600840336,\t0.050420168,\t0.147058824,\t0.613445378)\n",
    "PPV <- c(0.375,\t0.102803738,\t0,\t0.0595,\t0.059371069,\t0.059294471,\t0.023668639,\t0.034151034,\t0.037668799,\t0.032723773,\t0,\t0.112730061,\t0.113236419,\t0.195652174,\t0.133971292,\t0.096361186,\t0.324324324,\t0.165876777,\t0.107828656)\n",
    "Model<- c(\"LR\" , \t\"LRU\" , \"\tLRO\" ,\t\"KNN\" ,\t\"KNNU\" , \t\"KNNO\" , \t\"NB\" , \t\"NBU\" , \t\"NBO\" ,\t\"NBS\" ,\t\"boost_prob\" ,\t\"boost_prob_over\" ,\t\"boost_prob_under\" ,\t\"bag_prob\" ,\t\"bag_prob_over\" ,\t\"bag_prob_under\" ,\t\"RF_prob\" ,\t\"RF_prob_over\" ,\t\"RF_prob_under\")\n",
    "\n",
    "## PLoting ppv anf sensitivity from all models cutoff = 0.5 from file ppv_sev.csv\n",
    "df1 <- data.frame(col1= Sensitivity, col2= PPV , col3= Model)\n",
    "\n",
    "ggplot(df1, aes(x=Sensitivity, y=PPV , color = Model , label = Model )) + \n",
    "  ##geom_point(aes(size=17.5))+\n",
    "  geom_point() +geom_label_repel(aes(label=Model),hjust=0, vjust=0)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###### 10. Business Profits associated with all our 18 models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warning message:\n",
      "\"Ignoring unknown aesthetics: z\""
     ]
    },
    {
     "data": {},
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0gAAANICAMAAADKOT/pAAAAclBMVEUAAAAAp/8Ate4AuBMA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\nwa0DTLMAAAAASUVORK5CYII=",
      "text/plain": [
       "plot without title"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "library(ggplot2)\n",
    "library(reshape2)\n",
    "\n",
    "cutoff <- seq(0 ,1 , length = 100)\n",
    "\n",
    "\n",
    "##Importing Probabilities for every model\n",
    "prob_df <- read.csv('C:\\\\Users\\\\Jayant\\\\Downloads\\\\probability_df.csv')\n",
    "\n",
    "## Function to calculate frofit for all cutoff values\n",
    "profit <- function(x){\n",
    "  \n",
    "  revenue <- numeric(100)\n",
    "  cost <- numeric(100)\n",
    "  \n",
    "  profit_table <- data.frame(\"cutoff\" = cutoff , \"revenue\" = revenue, \"cost\" = cost)\n",
    "  \n",
    "  actual <- prob_df$Actual\n",
    "  predict <- x\n",
    "  \n",
    "  c_s_cost <- 0.1\n",
    "  revenue <- 1\n",
    "  \n",
    "  for(i in 1:100)\n",
    "  {\n",
    "    profit_table$revenue[i] <- sum(predict > cutoff[i] & actual == \"1\")\n",
    "    profit_table$cost[i] <- sum(predict > cutoff[i] )\n",
    "  }\n",
    "  \n",
    "  profit_table$profit <-revenue*profit_table$revenue  - c_s_cost*profit_table$cost\n",
    "  \n",
    "  return(profit_table$profit)\n",
    "  \n",
    "}\n",
    "\n",
    "\n",
    "profit_df <- data.frame(\"cutoff\" = cutoff)\n",
    "\n",
    "## Logistic Regression\n",
    "profit_df$Logistic_regression <- profit(prob_df$LR)\n",
    "profit_df$Logistic_regression_UnderSampled <- profit(prob_df$LRU)\n",
    "profit_df$Logistic_regression_OverSampled <- profit(prob_df$LRO)\n",
    "\n",
    "## K-Nearest Neighbour\n",
    "profit_df$KNN <- profit(prob_df$KNN)\n",
    "profit_df$KNN_UnderSampled <- profit(prob_df$KNNU)\n",
    "profit_df$KNN_OverSampled <- profit(prob_df$KNNO)\n",
    "\n",
    "## Naive Bayes\n",
    "profit_df$N.Bayes <- profit(prob_df$NB)\n",
    "profit_df$N.Bayes_UnderSampled <- profit(prob_df$NBU)\n",
    "profit_df$N.Bayes_OverSampled <- profit(prob_df$NBO)\n",
    "profit_df$N.Bayes_Smote <- profit(prob_df$NBS)\n",
    "\n",
    "## Bagging\n",
    "profit_df$Bagging <- profit(prob_df$bag_prob)\n",
    "profit_df$Bagging_UnderSample <- profit(prob_df$bag_prob_under)\n",
    "profit_df$Bagging_OverSample <- profit(prob_df$bag_prob_over)\n",
    "\n",
    "## Random Forest\n",
    "profit_df$RF <- profit(prob_df$RF_prob)\n",
    "profit_df$RF_UnderSample <- profit(prob_df$RF_prob_under)\n",
    "profit_df$RF_OverSample <- profit(prob_df$RF_prob_over)\n",
    "\n",
    "## boosting\n",
    "profit_df$Boost <- profit(prob_df$boost_prob)\n",
    "profit_df$Boost_UnderSample <- profit(prob_df$RF_prob_under)\n",
    "profit_df$Boost_OverSample <- profit(prob_df$RF_prob_over)\n",
    "\n",
    "chart_data <- melt(profit_df, id='cutoff')\n",
    "names(chart_data) <- c('x', 'func', 'value')\n",
    "\n",
    "ggplot() +\n",
    "  geom_line(data = chart_data, aes(x = x, y = value , z = x, color = func), size = 1)+\n",
    "  xlab(\"Model\") +\n",
    "  ylab(\"Profit\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "R",
   "language": "R",
   "name": "ir"
  },
  "language_info": {
   "codemirror_mode": "r",
   "file_extension": ".r",
   "mimetype": "text/x-r-source",
   "name": "R",
   "pygments_lexer": "r",
   "version": "3.3.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}