KMeans, DBSCAN, SpectralClustering, Hierarchical Clustering, MeanShift.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 基于sklearn的降维和聚类算法\n",
    "\n",
    "## ~~线性降维方法：~~\n",
    "* ~~FA(Factor Analysis)~~\n",
    "* ~~PCA(Principal Component Analysis)~~\n",
    "* ~~LDA(Linear Discriminant Analysis)~~\n",
    "\n",
    "## ~~非线性降维方法：~~\n",
    "* ~~KernelPCA~~\n",
    "* ~~LLE(Locally Linear Embedding)~~\n",
    "* ~~t-SNE(t-Distributed Stochastic Neighbor Embedding)~~\n",
    "\n",
    "<font color=\"#ff0000\">**这篇只讲聚类**</font>\n",
    "## 聚类方法：\n",
    "* K-means\n",
    "* DBSCAN——基于密度的空间聚类算法\n",
    "* Spectral Clustering——谱聚类\n",
    "* Hierarchical Clustering——层次聚类\n",
    "* MeanShift——均值迁移\n",
    "* Birch\n",
    "\n",
    "## 总结\n",
    "1. 只实现了K-means、Hierarchical Clustering、MeanShift、Birch\n",
    "2. 其中只有K-means各个标签对齐，其他几个算法MeanShift没调参比kmeans差，DBSCAN能跑调不好，层次聚类和Birch准确率高但是最后一维标签对不上"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-03-04T07:09:18.959235Z",
     "start_time": "2019-03-04T07:09:18.951239Z"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import seaborn as sns;sns.set()\n",
    "\n",
    "from sklearn.datasets import load_digits\n",
    "from sklearn.preprocessing import StandardScaler, MinMaxScaler  # 0-1化\n",
    "from sklearn.cluster import KMeans, DBSCAN, SpectralClustering, AgglomerativeClustering, MeanShift, Birch\n",
    "from scipy.stats import mode\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix  # 计算准确率和混淆矩阵\n",
    "\n",
    "from matplotlib import pyplot as plt\n",
    "from mpl_toolkits.mplot3d import Axes3D\n",
    "from munkres import Munkres\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-03-04T06:32:28.615931Z",
     "start_time": "2019-03-04T06:32:28.461057Z"
    }
   },
   "outputs": [],
   "source": [
    "dx = load_digits()   #读取数字数据，介绍如下\n",
    "''' \n",
    "    Each datapoint is a 8x8 image of a digit.\n",
    "    \n",
    "    =================   ==============\n",
    "    Classes                         10\n",
    "    Samples per class             ~180\n",
    "    Samples total                 1797\n",
    "    Dimensionality                  64\n",
    "    Features             integers 0-16\n",
    "    =================   ==============\n",
    "'''\n",
    "xx = dx.data         #数据\n",
    "xx = MinMaxScaler().fit_transform(xx)\n",
    "yy = dx.target       #标签\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-03-04T07:11:20.157977Z",
     "start_time": "2019-03-04T07:11:20.150981Z"
    }
   },
   "outputs": [],
   "source": [
    "# 对齐聚类标签和实际标签\n",
    "def predict_transfer(label, predict):\n",
    "    # L1 should be the labels and L2 should be the clustering number we got\n",
    "    L1 = np.array(label)\n",
    "    L2 = np.array(predict)\n",
    "    Label1 = np.unique(L1)\n",
    "    nClass1 = len(Label1)\n",
    "    Label2 = np.unique(L2)\n",
    "    nClass2 = len(Label2)\n",
    "    nClass = np.maximum(nClass1,nClass2)\n",
    "    G = np.zeros((nClass,nClass))\n",
    "    for i in range(nClass1):\n",
    "        ind_cla1 = L1 == Label1[i]\n",
    "        ind_cla1 = ind_cla1.astype(float)\n",
    "        for j in range(nClass2):\n",
    "            ind_cla2 = L2 == Label2[j]\n",
    "            ind_cla2 = ind_cla2.astype(float)\n",
    "            G[i,j] = np.sum(ind_cla2 * ind_cla1)\n",
    "    m = Munkres()\n",
    "    index = m.compute(-G.T)\n",
    "    index = np.array(index)\n",
    "    c = index[:,1]\n",
    "    newL2 = np.zeros(L2.shape)\n",
    "    for i in range(nClass2):\n",
    "        newL2[L2 == Label2[i]] = Label1[c[i]]\n",
    "    return newL2\n",
    "\n",
    "# 混淆矩阵绘图\n",
    "def draw_confusionmatix(label, predict):\n",
    "    sns.heatmap(confusion_matrix(label, predict).T, square=True, annot=True, fmt='d', cbar=False,cmap = 'rainbow')\n",
    "    plt.xlabel('true label')\n",
    "    plt.ylabel('predicted label')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### K-means"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-03-04T06:32:56.477481Z",
     "start_time": "2019-03-04T06:32:56.301797Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 199 ms\n",
      "accuracy_score: 0.7952142459654981\n"
     ]
    }
   ],
   "source": [
    "# K-means\n",
    "kmeans = KMeans(n_clusters = 10)\n",
    "%time yyk = kmeans.fit_predict(xx)\n",
    "#yyk = kmeans.predict(xx)\n",
    "\n",
    "yyk = predict_transfer(yy, yyk)\n",
    "acc = accuracy_score(yy,yyk)\n",
    "print('accuracy_score:', acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-03-04T06:33:01.948908Z",
     "start_time": "2019-03-04T06:33:01.471531Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x216 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#可视化10类中的中心点——最具有代表性的10个数字\n",
    "fig, ax = plt.subplots(2, 5, figsize=(8, 3))\n",
    "centers = kmeans.cluster_centers_.reshape(10, 8, 8)\n",
    "for axi, center in zip(ax.flat, centers):\n",
    "    axi.set(xticks=[], yticks=[])\n",
    "    axi.imshow(center, interpolation='nearest', cmap=plt.cm.binary)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-03-04T06:33:04.029454Z",
     "start_time": "2019-03-04T06:33:03.454267Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXIAAAGHCAYAAACte9mjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzt3XmclvP+x/HXPfvUNNoUTUol3xBZWqhQUlGWdOogWyXZl8IRThLHcg5xyBpSnFM6JFJKhxaE/Dq2qL60KUX7YqZmv39/XPeYkWaaaq5t5v18POZxX9d13/f1+XTd1/3pe3/v6/5+I9FoFBERCa84vxMQEZEDo0IuIhJyKuQiIiGnQi4iEnIq5CIiIadCLiIScgl+JyACYIy5FxgBDLDWjtvD/YcBHwGNgcestbeWY18Aw621fyvjsU8CN8ZWm1hrV+1H+iK+UotcAs8YcwjwAU4Rf6KsIr4HvcvYb6Ss+0XCQoVcAs0YUwd4H2gOPGWtvWUfnv4LcIIx5vBS7j8FyAAyDyhJEZ+pkEtgGWMOAmYBxwDPWmtv3MtTdvdW7PaCUu7vA2zH6bIRCS31kUsgGWOqA+8CJwJjgOv3YzezgQtxuk8e38P9vYGpQHopOZwI3AOcClQDLPAc8Ly1NrrbY8+J5dgaqAlsA+YD91prvyrxuLnA4UBH4B9AdyAVWAjcY62dW+KxCcDdwJ+AI4Bs4P+Af1hrPyjvQZDKTy1yCRxjTApOgW0PvARcs3vhLKc84B2gvTGm/m4x2uL0ub9eSg5nA58AZ8T2MRrn/fIs8Pxuj70h9pjmwETgn8Bi4HzgQ2PMobvtPg3nU0ArYDzOJ4cOwHvGmGYlHjcauBfYElv+D9Au9rhO5TsEUhWokEvQJAKTcQoowPv7WcSLTMY5z8/fbXsfYAdO183vGGOq4RTYHcBx1tr+1tq/AMcDbwBXGWN6xB6bDDwAfA+0stbeZK293Vp7Ok7RrwGcu1uIOsDXwPHW2qHW2n7AcCAJuCK233RgMPChtbaTtfYOa+01wJlAPPv3CUUqKRVyCZr7gR7ATKAQeMYY0/AA9jcL58vM3a9O+RMw1Vqbs4fnnAccjNOFsapoo7W2ELgztjogdhsPXAUMstZm7bafubHbenuIMcpam1di/d3Y7ZGx2zggAjSKXXpZlMNCoBnQbw/7lCpKfeQSNPWBV4H+wCPAUGC8MebM/WmZW2uzjTHTgd7GmIOstdtjfd9NgSGlPO2kotvYNem7K8BpnWOt3YnT5YEx5kjgaJxC2xLoEnt8/B728f1u69tjt8mx/W4zxkwCLgKWG2PmAzOAadbaxWX8k6UKUiGXoJmM86OgQmPM3UBPnG6WocCoogeVUmDfKvnF4m77vBA4B/g3TrfKr8B7peRQM3Z7URl51i6Ry2k4X6aeGNuUjdN18j/gMJyW9e52/yRQ9J9UycdejvMl6ACgU+zv78aYhcBVpfxbpQpSIZegmWatLYDfWtP9gY+BB4wxs6y1i2KPG7GH564C9lTc3gV24XSv/BunW+WdUrpVoPi68i7W2tllJWuMaYzTUs7G6dP+GPjeWltgjLkQ6FXW88sS63oZBYwyxjQCugJ/BroB04wxTXbrnpEqSoVcAs1a+5kxZhTwF+Dfxpg21toca+2eWrml7SPLGPMecJYxpg1OP/QdZTzlm9hta5xLGH9jjKmNc0niQmvtv3AKdTXgL9baF3bbz1Gx23LnWiJOE5y+90+stdOstatxruB5yRjzAc6nlCb8sYtGqiB92SlhcA+wBDgWeGg/9zEZp+A+idPinlnGY6fgXLFyR6zfu6R/ADfjXNcNTkscnL793xhjjos9DpwrcfbVLpz/bO6PXRlTtN8k4FCcrplf9mO/UgmpRS6BZ63NMcZcAXwK3GKMmb4fP4h5B8gFTgYmWmuzS3tg7IvGQcAE4EtjzBRgHU4fdRucH+U8Gnv4NJwf/9xljGkBLMe5nvwcir/ArLOPuWKt/cUY80+c7wa+jX1hWwichdPSv99au2Nf9yuVk1rkEgrW2v/DaQ1HcK5iqbWPz9+OM/AWONeC7+3xrwOnxZ5zNs4IiTVwLo8801qbGXvcWpxru2fjXKVyHU7XzZNAC2AzTpfOPnev4HQnXYvz6aA/Th/8r0B/a+09+7E/qaQi0eiB/NZCRET8pha5iEjIqZCLiIScCrmISMipkIuIhJwKuYhIyAXzOvJIxNdLaUbi/5U8hXsaZsljcQV+ZyA6D6TIiGjpvxBWi1xEJORUyEVEQk6FXEQk5FTIRURCToVcRCTkVMhFREJOhVxEJORUyEVEQk6FXEQk5FTIRURCToVcRCTkwl3I27aFOXOc5YkTneU5c2DlSme9e/fibXPnQn4+tGjhbk4R6PksDPwErpgDtZq5G640GW2h/77OallRgnAM/M7B7/gl+HYuBOEYVJEcgjloVnncfjtcdhlkZTnrF1/s3Nas6RTuIUPgl1/gvfec7bfdBvPnw9KlrqbVohckpMDY9pDRDrqNgkm9XA35Bx1ug1aXQO5Ob+MWCcIx8DsHv+MX8fNcCMIxqCo5uN4iN8a4E2P5cujd+4/bR46E0aOdIl4kI8Mp+iNHupJKSY06wrKZzvLaBdCgtesh/2DLcnitr/dxiwThGPidg9/xi/h5LgThGFSVHFxpkRtjmgKPAa2B/FgxXwQMsdZ+XyFB3nwTGjf+/baDD4YuXZzWeElDh8Ljj0NuboWELktyOuRsL16PFkAk3rn1ypIpULPx3h/nliAcA79z8Dt+ET/PhSAcg6qSg1tdKy8Cd1prFxRtMMacDLwMdHApJvTpAxMmQGFh8bZIBM45B+6+27WwJeXsgKQaJcLHef/m9VsQjoHfOfgdPwiCcAyqSg5uda2klCziANbaz1yKVezMM2HGjN9va9nS6RfPznY9PMCa+dC8h7Oc0Q7WL/IkbKAE4Rj4nYPf8YMgCMegquTgVov8a2PMWGAmsB2oAfQAvnEpnsMYWLFi79tctGQKNO0KA+cDEXh7gGehAyMIx8DvHPyOHwRBOAZVJYdINFrx05oZYyJAL6AjkA7sAOYDU6y1ew+oqd40xZcAOg+kWFlTvbnSIo8V6ymxPxERcVG4fxAkIiIq5CIiYadCLiIScirkIiIhp0IuIhJyKuQiIiGnQi4iEnIq5CIiIadCLiIScirkIiIhp0IuIhJygZzqze9Bq15etNrX+ACXndTI7xR8Hywpu8beH+O2lF/9zsB/+Ul+ZwCZdfyNv/TUfH8TAMoq12qRi4iEnAq5iEjIqZCLiIScCrmISMipkIuIhJwKuYhIyKmQi4iEnAq5iEjIqZCLiIScCrmISMipkIuIhJwKuYhIyAVy0Kz9FoGez0D9VlCQA1MHwdbl3oROvO0aqFYdgGi9QyhsfTIJr7xAtO7BAORfeDnRY1q5nkdcAlwwBmo2hoRkmPcwLJ3metjf8/F1+C2FOOg9Guo2h2gBvHEdbFnpZQL+H4MiGW2h60Mwrou3cQNxLgJdboCW3SA+CeaPhwUT3Y8ZHw9Dromj/sEREhNh4puFrP4pyq3XxRONwo9rojw9tpBoBY0PWKkKeYtekJACY9tDRjvoNgom9fIgcG4uAHn3jfptU/yEl8m/7CoKTznVgwSKteoHO7fA5IGQWhuuW+D9m8e316GEo852bp/vBk06Qs8H4dWLvYsfhGMA0OE2aHUJ5O70PnYQzsVmp0CTNvDk+ZCYCp2v9SbuGadG+DUTHn26gBpp8NTf41mxKsr4SYUsWhzlhkFxnNI6wif/VzGVvFIV8kYdYdlMZ3ntAmjQ2pu4kVXLISeHxPvugIJC8i8ZQNyKH4iuXEb89DcpPMJQcNlVzn/TLvtuMnz3ZvF6oQ9D0fr1OpS0eDosjeVQqxFkbvA2fhCOAcCW5fBaX+g93vvYQTgXW3SCdUtgwFhISYN37vcm7kefRvn4s+IiXVAARzSNsGhxIQALv4py4nEVV8grVR95cjrkbC9ejxZAxP3aCckpFJzXl7zhD5N/9c0k/vNhClu2Iv/KG8i7/zEi2dnEvfeOB4lAbhbkZkJSGlw0ET4Y4UnY3/HtddhNYQH0fQ7O/Qd8+7a3sYNyDJZMgcI87+NCMM7F6rWhUSsYPxheHwaXPu1N3Owc2JUNqSlw99A4XplUSKTE/bt2QfVqFRfPlRa5MWYOkLzb5ggQtda2dyMmQM4OSCoxGUEkznkDuS3aIIPoIQ0gEiHaoCHRGukUdOwMdesBUNjmFOI++5hC91MBIL0h9PsPfP48fDPJo6Al+PU67Mnr10BaPbhuNjzeFvI86mII0jHwk9/n4s6tsGEZFOTBxuWQlw1pdSBzs/ux69aB4bfGM31WIXPnR7nykuL7UlMhM6viYrnVIh8GpAGXARfH/i6K3bpmzXxo3sNZzmgH6xe5Ga1Y3AfvkTD+eWdlyybYmUXSsBth80YAIou+pLBZc09yqV4P+k+HWXfDFz58nAb/XoeSTrgITh/qLOftgmiht4U0CMfAb0E4F1d8Di06O8vp9SGpGmRtdT9uzYPggbvjeXlCIbPmOt0ny1dFOfZop13e+vgI3y2tuJnQItGK+tp0N8aY24Fl1top+/rckZH9nOut6EqB45zltwfAZrvvu9nnqd7y8kh46hEimzZAJEL+pYMgexcJE8dBUhLRho3Jv/J6SCj/B6D9neqtxyho2Qc2lfh3v3Ie5Gfv+74ScvcrhQp7HQ5kqrfEatDnGahRH+ITYe5jsOTdfd/Pfk/1VkHHoLACumNqNoY+E+DFDvv3/P3NoSLPxQOZ6u3cv8IR7Z1PRdMfAjtv3/exr1O9XX1FHKe1j/DT2uJtz40v4Nr+8SQkwJq1UZ54vpDCfah0MyYlREq7z7VCfiD2u5BXEM3Z6djvQl5BNGdnxRTyypCD5uwsu5BXqi87RUSqIhVyEZGQUyEXEQk5FXIRkZBTIRcRCTkVchGRkFMhFxEJORVyEZGQUyEXEQk5FXIRkZBTIRcRCTmNtbIH71/j8yAjwEd1uvudAiMenuNr/LgqOOyr7Jnf4+74PeYOwIgoGmtFRKSyUiEXEQk5FXIRkZBTIRcRCTkVchGRkFMhFxEJORVyEZGQUyEXEQk5FXIRkZBTIRcRCTkVchGRkFMhFxEJuQS/E6hQEej5DNRvBQU5MHUQbF3uftj4OLjz8ngOqRMhMQFemVHA/G+ccb/ObBPHnzrFce0j+e4Eb3gUdB8MLw0p3nZcFzj5Ahhzg7Pe4c/OtmghzPs3LPnYnVx2k9EWuj4E47p4Eu73fDoXAhM/CDn4HR+IxEHv0VC3OUQL4I3rYMtKb3Pw4jhUqhZ5i16QkAJj28P7w6DbKG/idm8Xx/YsuGFUPrc/lc+QC53/H49oGOGc9nFESh2z7AB1vAh63QYJScXbDmkGJ/Xgt6Ap1eGU3jDmehh3O/S83qVkfq/DbXD+887r4Qe/zoWgxA9CDn7HBzjqbOf2+W7w3weg54Pe5+DFcfC0kBtjkt3cf6OOsGyms7x2ATRo7Wa0YnO+KOTFqcVjrhYUQnp1uKZXPE++7uJYrFvWwcR7itdT06HbYHj3qeJtudmwbT0kpkBSCng0bPGW5fBaX09C7ZFf50JQ4gchB7/jAyyeDlNucpZrNYLMDd7n4MVxcKVrxRhzLvAUkAfcba2dFLtrBnCGGzEBktMhZ3vxerQAIvHOrZt25Ti3qclw/1UJvDi1gGGXJjD6jXxy3BzafPGHULO+sxyJgwtuhxlPQ17O7x+3fSPcPM55zIcTXEyo2JIpULOxJ6H2yK9zISjxg5CD3/GLFBZA3+fg6HNgwuXexgZvjoNbfeR3AycAEeB1Y0yKtXZ8bN01OTsgqcQA9JE4706aerXggasTmDKvkJ82RGlYD269OIGkRDj8kAg39o1ntJut8wZHQp2GcN4Qp6vl4MbQ43pY8SXUqA2jLnYed8Uj8OO3sHape7kEgJ/nQhDiByEHv+OX9Po1kFYPrpsNj7eFvJ3exfbiOLjVtZJrrd1ird0MnA/cYIzpDO7O/LNmPjTv4SxntIP1i9yMVqxWDRh1YyLPTSng3U8LWfJjlMvvz+emx/O598V8Vv0SdbeIg1OYRw9wvvScdB9s/BHefRp2/eq00PPznL/sTEhNczeXAPDrXAhK/CDk4Hd8gBMugtOHOst5u5zv+73+z8SL4+BWi3yVMeYxYLi19ldjTG/gPaCmS/EA5+N8064wcD4QgbcHuBmt2GVnxVOjGlzRI54rYi/YbU/lk5vnTfwy/bgI1lq4+hnnLP5xESxb6HdWrvPrXAhK/CDk4Hd8gG+nQp9nYPAMiE+EacMgP2fvz6tIXhwHV+bsNMYkAJcC/7HW7oxtqw/caa29ZW/P15ydmrMTNGenFNOcnWXP2elKi9xamw+M223bemCvRVxERPZNpbqOXESkKlIhFxEJORVyEZGQUyEXEQk5FXIRkZBTIRcRCTkVchGRkFMhFxEJORVyEZGQUyEXEQk5FXIRkZBzZdCsA+X3oFmF8X5Gd6xoU+h3Crx6xIm+xh/5r698jQ/+nwsaOMyx09VxU/cuN9Xf+ACPryt90Cy1yEVEQk6FXEQk5FTIRURCToVcRCTkVMhFREJOhVxEJORUyEVEQk6FXEQk5FTIRURCToVcRCTkVMhFREJOhVxEJOQqVyGPQM9nYeAncMUcqNXMnzQy2kL/D7yP2+xouOtpZ1ydBofDX5+LMPz5CFfcFiHi5ivdrCXc9aKz3LgFPDHLWb/rRWjXzdl+6nlw76sw8t9w/lUuJhNTxc8FwP9j4Hf8EqrXhTu/hYOb+xO/yw1w81QYOhPaXVzx+0+o+F3+kTEmFSiw1ua6GadFL0hIgbHtIaMddBsFk3q5GfGPOtwGrS6B3J3exu15CXQ4O0LOLmf9z9dEeP25KPYrGPzXCCeeCv+b50bg/tChJ78FPvwomPkqzHi1+DH1GkKXvvDAIMjPhd7XQnwCFOS7kJCjKp8LRfw+Bn7HLxKXAL0fh7xd3scGaHYKNGkDT54PianQ+dqKj+FKO80Y08QY85Yx5jljzJnAEmCJMeYcN+IVadQRls10ltcugAat3Yy2Z1uWw2t9vY+7fi08Max49N8n7nKKeHwCHFQHdmxxK/AaeOLW4vUmR8Hxp8LdL8GgEZBSDY5pBysXw9X3O9u//8rVIg5V+1wo4vcx8Dt+kZ73w4KXYccv/sRv0QnWLYEBY2HQeFj834qP4dYH7peBx4FPgTeAtsAJwJ0uxQMgOR1ythevRwsg4vF40kumQGGetzEBFs79fW2MFkKdQ+DhCRHSDoKff3Qr8Ae/D7z8O5j4ODxwJWxYCxdcDTVqgTkRXrwXnrwVLh8G1Wq4lJCjKp8LRfw+Bn7HBzjpYsjaBN/P9jZuSdVrQ6NWMH4wvD4MLn264mO4VcgTrLXzrLXjgbestRustTsAV5thOTsgqUR9iMQ5J09VtfkXuP3PUWZPidLv5lLHpK9Y/5sNq5Y4ywtnO33mmdtgyULI3gk7tsLa5XBIY1fT0Lng/zHwOz5A60uheWcY/A40OBYufBbS6nmbw86tsHQuFOTBxuWQlw1pdSo2hluF3BpjXjTGxFlr+wMYY4YBrn64WTMfmvdwljPawfpFbkYLtiH/iFC/obOcvdNpoXviL89A05bO8jFtnaL+/VdwVGtITILkFMhoButXu5qGzgX/j4Hf8QGe7wnPnwNjzoV1i2DStZC5wdscVnwOLTo7y+n1IakaZG2t2Bhufdl5FXCutbZk+fgJeNKleIDzUbZpVxg4H4jA2wPcjBZs016NMnh4hPw8yM2GFx/yaPa8lx+AK+6E/DzYvgleuh+ys2DeWzB8HEQi8NYYyNrhaho6F/w/Bn7HD4rF70Ozk2HIu86nksl3VXzDSnN27oHf8zSC5uwEzdkJmrOziObs1JydIiKVmgq5iEjIqZCLiIScCrmISMipkIuIhJwKuYhIyKmQi4iEnAq5iEjIqZCLiIScCrmISMiVOtaKMea0sp5orf2w4tMREZF9VdagWSPLuC8KnFHBuYiIyH4I5KBZIxL8HTRLAxUFw4iW9f1OgeHfr/c1foKrkyNKmIyIlj5o1l6HsTXGNAZeBA4HTgUmAAOttasqKD8RETkA5fmy83ngESATWA9MBF5xMykRESm/8hTyutbaWQDW2qi19gUg3d20RESkvMpTyHcZYxrifMGJMaYjkONqViIiUm7lmeptCDANaGaM+QqoDfR1NSsRESm3vRZya+1CY0wb4EicFry11uq7dBGRgNhr14ox5iDgYZwvOF8AhhtjqrmdmIiIlE95+shfAgqA/sDVQA1gjIs5iYjIPihPH/kR1to+JdZvMcZ841ZCIiKyb8rTIrfGmPZFK8aYVsAP7qUkIiL7oqxBs1biXHKYCvQxxizF6WI5ChVyEZHAKKtrpZNXSVS0jLbQ9SEY18WH4BHo+QzUbwUFOTB1EGxdXoXi+5XDsSfCkL/CwN5Quy7cOwrSD4K4eLjrBqieBnfcX/z4406Cm/vD/DmupBOXABeMgZqNISEZ5j0MS6e5Eqp0fp8LfsevQjmUWsittT8CGGOSgR5AmpMS8UAT4J6KTaVidLgNWl0CuTv9id+iFySkwNj2kNEOuo2CSb2qTnxfchhwPZzbF3bGXvShw2H6ZHhvKrTpAE2aw0fvO0UeoNu5sHG9a0UcoFU/2LkFJg+E1Npw3QLvC7nf54Lf8atSDuXpI58I3AQ8CJwVuz2qvAGMMfX2L7X9s2U5vObjz5UadYRlM53ltQugQeuqFd+XHNasglsGFK8f3xbqHwovvA49/wQLPym+L7UaXHc7PHS3qyl9Nxk+uLd4vdCHETX9Phf8jl+VcihPIT8OZ+zxKcA/gA44IyHukTHmyJJ/wNQSy65bMgUK87yItGfJ6ZCzvXg9WgCR+KoT35cc3p8O+fnF6w0Ogx3b4aq+8MtPMPCG4vsu6Aez3oFtW1xMCHKzIDcTktLgoonwwQhXw+2R3+eC3/GrUg7lKeQbrLVRYClwnLV2BZBUxuPfB6YCz+GMnGhit88dYK6hkLMDkmoUr0finBeuqsQPRA7bt8Kc95zlubPgmFbF9/X8E0z+tydppDeEgbPg6wnwzSRPQv6O36+D3/GrUg7lKeTfGmNGA3OBIcaYYVD6AOdAa2Ax8JC1tjPwlbW2s7W2SswotGY+NO/hLGe0g/WLqlb8QOTwxQI4NfZN90mnwDLrLKfVgKQkWL/O9RSq14P+02HW3fDFeNfD7ZHfr4Pf8atSDuX5QdC1QHtr7WJjzD3AmUC/0h5srd1gjPkz8GhsjJYqZckUaNoVBs4HIvD2gL0+pVLFD0QOj94LIx+DC/tD5g6441pne+NmsG6NJymcfgek1IROdzp/AK+cB/nZnoQH/H8d/I5flXIodaq3iph82RjTHxhgrT19X5LSVG8CmuoNNNWbFNvfqd4OePJla+04YNzeHiciIvuvrOvIO3uZiIiI7J/yfNkpIiIBpkIuIhJyKuQiIiFX1uiHc6D0q0eqynXhIiJBV9ZVK/fGbq8CdgHjgXzgYpyhbUVEJADKumplHoAx5lFrbckf9nxmjFnoemYiIlIu5ekjTy054JUx5lgg0b2URERkX5TnJ/pDgbnGmLU4hb8eTveKiIgEwF4LubV2ljHmcOBYnC8/v7HW5pf9LBER8cpeu1aMMbWAp4FHgFXAmNg2EREJgPJ0rbwAzALaApnAz8C/gJ5uJeX3oFWFHg88vyd+HwOA/LJGnffAyG/9HbAK4P5Jl/oaf+SF//I1PsAOT+f4CqYg1ISylOfLzibW2jFAobU211p7N9DQ5bxERKScylPI840xBxH7cZAxpjlQ6GpWIiJSbuXpWhmBMztQI2PMW8ApwEA3kxIRkfIrz1UrM2M/AGoHxANXA1vdTkxERMpnr4XcGPOptfYUYHpsPQ74GudyRBER8VlZg2bNBjrFlgspHkCrAJjqemYiIlIuZY21cgaAMeYJa+3N3qUkIiL7ojxXrbxojHkNwBhzlDHmQ2OMcTkvEREpp/IU8hdwhrDFWrsEuB94yc2kRESk/MpTyKtba2cUrVhr/wtUdy8lERHZF+W5jnyDMeYanJ/lA1wE+P/baRERAcrXIh8AnIMzxspqnDFWBrmZlIiIlF95fhC0GqeQB18Eej4D9VtBQQ5MHQRbl3ufRkZb6PoQjOvifewgHIO4BLhgDNRsDAnJMO9hWDrN2xy8Pg5fr8nk0Vk/8eqVLVi2YRfD315FFGhxSCrDezYmPi7CvO+38fScdQAc3aAaI85pTCQScS+pAJwLAEPfh107nOUtq2HSLVUrPkCXG6BlN4hPgvnjYcHEit1/WdeRT7PWnmOMWckeJmG21jat2FQOXItekJACY9tDRjvoNgom9fI2hw63QatLIHent3GLBOEYtOoHO7fA5IGQWhuuW+B9IffyOLzw0c9M/WozqUnOB9zH/vsTQ7s2pM3hNRj25gpmL93GKc3SeeS9n3hloKF29URe+Ohntu7Mp3Z19ybbCsK5kJDs3D7b29u4QYkP0OwUaNIGnjwfElOh87UVH6OsFvlVsdtOBxIg9kvQQ4GfrbWuDrbVqCMsm+ksr10ADVq7GW3PtiyH1/pC7/Hex4ZgHIPvJsN3bxavF/owJK+Xx6FR7WRG9zuCv7yxAoDRFx9BfFyE3PxCNv6aR520BL5cncmR9VP5+8w1rNmaQ9+TDna1iEMwzoUGxzjFa/Ak55Pauw/C6v9VnfgALTrBuiUwYCykpME791d8jLL6yLsaYy4HTi/lr1TGmJdit+2A74E3gW+NMSdXRNKlSU6HnO3F69ECiHg8jvCSKVCY523MkoJwDHKzIDcTktLgoonwwQhv44O3x6H7MbVJiCvuIomPi7B2Ww7njP6WrTvzaVI3la0781mwcge3dTuMFy47kvGfrGflpmx3EooJxLmwC+Y+C2MuhDduh0uegTgPc/A7PkD12tCoFYwfDK8Pg0ufrvgYZbXIO8dumwFHAO8C+cBZwHfAK2U8t0ns9gHgbGvtD8aYBsBE9vKfwIHI2QFJNYrXI3HOyVuVBOUYpDeEfv+Bz5+HbyZ5H9/v45CnUK8dAAAboklEQVRRM5lZQ47j9YUbeXjGanoeV4djM6pzcA2nFd768DSW/LyTJnVTXMvB72MAsHE5bFrpLG9aATu3Qnp92LauasQHJ+aGZVCQ5+STlw1pdSBzc8XFKLVFbq0dYK0dgDO2ynHW2kHW2muAE4C0cu6/wFr7Q2x/68qKVxHWzIfmPZzljHawfpGb0YIpCMegej3oPx1m3Q1f+NTF5OdxuOZfP7Bqs9Parp4cR1wkQssG1fh+/S62ZOWRXxDl65+yOKKee0UcgnEutL0YzrvXWU6v73Qt7PDw4mW/4wOs+BxadC7OIakaZFXw+LHluY68AbClxHoWTp93WWoaY/4HVDfGXAn8GxgF/LhfWZbTkinQtCsMnA9E4O0BbkYLpiAcg9PvgJSa0OlO5w/glfMg392ehN/x8zgMPu0Qhr25ksT4CKmJcfyt1+HUrp7IrV0bMuiV7wE465jaHFm/mqt5BOFc+HwCXPQk3DAVolHnihEvvzPxOz7A4veh2ckw5F3nU9HkuyBawd8WRqLRP1yQ8jvGmMeAVjj93BHgz8A8a+3wvTwvOfa8nTj95AOBl6y1e+1BHhn541UyXgrC/HyasxMScv2NDzBCc3Zqzk6CURMeX0ep16qW5zryocaYP+FcvRIFHrXW7nUYW2ttDvB5iU3P7T1VERHZV+Xts/4F5wvO24Ft7qUjIiL7aq+F3BhzM/A3YCjOYFnPG2NuczsxEREpn/K0yPsD3YEsa+0WoA2afFlEJDDKU8gLrLUlv3bKxrkkUUREAqA8hXyeMeZRnEsJe+HM1/mBu2mJiEh5laeQ3w78AHwNXI7zC0/1kYuIBER5fhA0w1rbHXje7WRERGTfladFXs0Yc5jrmYiIyH4pT4u8LrDKGLMB2IXz685oEMcjFxGpispTyM9yPQsREdlv5Snk64DrgTNwhrF9F3jJzaRERKT8ylPIXwRSgRdw+tQvB1oCPsx8540gDJAThEGzgjBold9G9PN30KqRpY+T5F0OG3wdww7QAG57U55C3s5a26JoxRjzDvCteymJiMi+KM9VKyuNMUeUWK8PrHUpHxER2UflaZEnAl8bYz7E6SPvCPxsjJkNYK09w8X8RERkL8pTyHef8/lRNxIREZH9U56JJeZ5kYiIiOwfVydDFhER96mQi4iEnAq5iEjIqZCLiIScCrmISMipkIuIhJwKuYhIyJXnB0HhEYGez0D9VlCQA1MHwdbl3qYQlwAXjIGajSEhGeY9DEuneZhAAI6BciiW0Ra6PgTjungQrG1b+PvfoXNnmDgRDjnE2X744fDZZzBuHAwb5myLRKBjR2jZEpYudSefALwGvr8fwZPj4EkhN8bUBTZba10dRq1FL0hIgbHtIaMddBsFk3q5GfGPWvWDnVtg8kBIrQ3XLfD2xAnCMVAOjg63QatLIHenB8Fuvx0uuwyyspz1iy92bmvWhDlzYMgQ+OUXeO89Z/ttt8H8+e4VcYLxGvj9fgRvjoMrhdwYMwA4DJgGTACycaaMu85a+74bMQEadYRlM53ltQugQWu3IpXuu8nw3ZvF64UeD0cbhGOgHBxblsNrfaH3eA+CLV8OvXvDq6/+fvvIkTB6tFPEi2RkOEW/TRtXUwrCa+D3+xG8OQ5u9ZFfB4wCHgHOs9YeD3QCHnIpHgDJ6ZCzvXg9WgARj8cWz82C3ExISoOLJsIHI7yNH4RjoBwcS6ZAYZ5Hwd58E/J2C3bwwdCli9OlUtLQofD445Dr7iDbQXgN/H4/gjfHwa1CnmetzQJ+BVYAWGvXAa52reTsgKQaxeuROOegeS29IQycBV9PgG8meRs7CMdAOQREnz4wYQIUFhZvi0TgnHPgtddcDx+U18DP9yN4cxzcKuRTjTFvA98B04wxQ4wx7wGzXYoHwJr50LyHs5zRDtYvcjPanlWvB/2nw6y74QsvPlLvJgjHQDkExJlnwowZv99W9OVmdrbr4YPwGvj9fgRvjoMrfeTW2oeNMacD3YHVQD3gSWvtdDfiFVkyBZp2hYHzgQi8PcDNaHt2+h2QUhM63en8AbxyHuS7/74BgnEMlENAGAMrVux9m0uC8Br4/X4Eb45DJBr1fz6+3Y2MuNsFszd+zw8IwZ8jsKrwe/7WkQUBmLPT37cj4P97MgjvxxHR0idw1Q+CRERCToVcRCTkVMhFREJOhVxEJORUyEVEQk6FXEQk5FTIRURCToVcRCTkVMhFREJOhVxEJORUyEVEQq5yTfVWQYIwrkJ2jb0/xvUc0vyN7/f4GgB1f/Q3fhDGORnRuJbfKXDnlq2+xt92qK/h90otchGRkFMhFxEJORVyEZGQUyEXEQk5FXIRkZBTIRcRCTkVchGRkFMhFxEJORVyEZGQUyEXEQk5FXIRkZBTIRcRCbnKNWhWBHo+A/VbQUEOTB0EW5dXrRwicdB7NNRtDtECeOM62LLSu/hFutwALbtBfBLMHw8LJnobv10f5w8gIRkaHg13t4FdOzxKoKqei8efBMPuhYvOhdEvwcH1nO0NG8GXC+HGK+Hmv0DnblBQAPfdCV9/4Vo6QXg/eHEuVqpC3qIXJKTA2PaQ0Q66jYJJvapWDked7dw+3w2adISeD8KrF3sXH6DZKdCkDTx5PiSmQudrvY0PsOAN5w+g733w2X88LOL4fx74ksPVN8EFF8KuLGf9xiud2/SD4LV34L674JjjoF0H6HUmNGgIz46H87u4llIQ3g9enIuudK0YY9Ld2O/eNOoIy2Y6y2sXQIPWVS+HxdNhyk3Ocq1GkLnB2/gALTrBuiUwYCwMGg+L/+t9DkUOOxYOPRI+8fgTgd/ngS85/LgSrrnsj9uH3AnjxsDG9dDmZPhojrN93U+QkAC167iWUhDeD0XcPBfd6iP/xRhzpUv7LlVyOuRsL16PFkAkvurlUFgAfZ+Dc/8B377tbWyA6rWhUSsYPxheHwaXPu19DkW6Xw8znvA+bhDOA89zmPkO5Of9fludutDhNHhjgrOeVgN+LdEczcyEGu62+/x+PxRx81x0q5B/DZxgjJltjDndpRh/kLMDkkpMyBCJc05eLwUhB4DXr4FRJ8IFT0JiNW9j79wKS+dCQR5sXA552ZDmXqOrVKnpUK8Z/PCp97GDcB4EIQfOPh/engyFhc565q9QvcSMJWlpsGP7np9bgfx8P4D756JbhXyXtfYG4C/ATcaYb40x/zTG3ORSPADWzIfmPZzljHawfpGb0YKZwwkXwelDneW8XRAt9P7Nu+JzaNHZWU6vD0nVIMuHCV6atQX7sfdxwf/zICg50PF0mFuib23hAjjtDIhEnD7yuDjYusW18EF4P4D756JbX3ZGAKy1C4E/GWMOAk4DjEvxAFgyBZp2hYHznQzeHuBmtGDm8O1U6PMMDJ4B8YkwbRjk53ibw+L3odnJMORdpxU4+S7nDeS1+k1h8xrv44L/50FQcqDpEbBmVfH6t1/D/30KU2Y5J8fw210NH4T3A7h/Lkai0YqfE9AYc4W1dvz+Pn9kJAATFfpMc3Zqzs6g0JydkFnb1/AAjF7lNJD3xJWulQMp4iIism/0y04RkZBTIRcRCTkVchGRkFMhFxEJORVyEZGQUyEXEQk5FXIRkZBTIRcRCTkVchGRkFMhFxEJORVyEZGQq1RTvVWUxrmv+p0C/9fiUr9TIK6g1DF6PFHN/WGqAy8IA4cN/9nfAasA2mx9wdf4X9a6ytf4e6MWuYhIyKmQi4iEnAq5iEjIqZCLiIScCrmISMipkIuIhJwKuYhIyKmQi4iEnAq5iEjIqZCLiIScCrmISMipkIuIhFzlGjQrAj2fgfqtoCAHpg6CrcvdD5ufm8/0v77Dtp+2kVQ9ie5/PZttP21l7uOzSUxNpGnHZnS4+lT3EwHiEuDyR6BOQygsgAl3wfoVnoT+Tbs+zh9AQjI0PBrubgO7dnibR/W6cNNcePEC2PiDt7H9OhdLikuAC8ZAzcbO6zDvYVg6rXLHX73oZ2Y+MY/BL17EptVbeWPETCIRqN+sLufdeSZxcc5AcLm78niu/wS633QapkMTV3Py4jhUqkLeohckpMDY9pDRDrqNgkm93I/71RtfklQtiSsmDGDzys28d/8MtqzaTL+XL6PWYbWYesdbrPliNYed2Mj1XFp2gvh4GNUXWnSAc2+FF693PezvLHjD+QPoex989h/vi3hcAvR+HPJ2eRu3iF/nYkmt+sHOLTB5IKTWhusWeFvIvY4/b9znfDl9MUmpiQC8O2ou3a7vQNPWjZjyt/+yZO4yjjmjOQBTH34fIt6M7unFcfCka8UYk2SMSXU7TqOOsGyms7x2ATRo7XZEx6blm2jasRkAdZrU4aev1pCSnkKtw2oB0PCEw/jpizWe5LJ+pVPEIhFISYOCfE/C7tFhx8KhR8InE72P3fN+WPAy7PjF+9jg37lY0neT4YN7i9cLCyp3/DoNa3Lpo+f/tr52yXqanHQYAKZDE5Yt+BGAD1/5Pxq1yuDQIw92N6EYL46DK4XcGHOkMeYNY8wEY8zJwLfAd8aYC92IVyQ5HXJKjGEdLYBIvJsRHfVb1GfZvB+IRqOs/fonCnILyMvOY/OKTRQWFLL8o2Xk7spzPxEgZyfUzoDh/4V+D8Lc8Z6E3aPu18OMJ7yPe9LFkLUJvp/tfewifp2LJeVmQW4mJKXBRRPhgxGVO37LM48kPrG4pEWjUSKxVndy9SSyM3NYtuBHNq/eStvex7mbTAleHAe3ulZeAO4HDgKmAa2AbcD7wCSXYpKzA5JqFK9H4pw3kNtaXXA8m1dsYsKAV2l4wmEccvShdL2zGzPve5eU9FRqH16HajWruZ8IcMYAWPIRTH0Uah4KN/8LHjgb8nM9Cf+b1HSo1wx++NTbuACtLwWicEQnaHAsXPgsjOsHmRu8y8Gvc3F36Q2h33/g8+fhG9feecGMH4kr7jrJycolpUYyC99axLafdzBm0GtsXLWFdUvXU6NudRqYeq7m4vZxcKuQJ1hr3zfGRIAHrbVrAYwxrjZL18yHI8+Fxa87/ZLrF7kZrdi6b9fR8MTDOPOObvz87Tq2rtnK8o+W0/eZi0hMTeTNm9/guAtaeZLLzh1QEDvKO7dBfALEedwSBGjWFuzH3scFeL5n8fLgd2DKUG+LOPh3LpZUvR70nw7TboEVc6pe/AYt6rFi4Wqatm6Enb+SZm0O47juLX67//V7ZnBc9xauF3EvjoNbhXyVMea12P4zjTEPANuBn12KB8CSKdC0KwycD0Tg7QFuRitWu3FtPnpqLgvGfUZKjRR63HcOyz/8gVcvG09CSgLH9GzJwUd40x83Zyxc+jAMeQ0SEp2Wea4PX/jVbwqbvflaIJD8OhdLOv0OSKkJne50/gBeOQ/ys6tG/B5DOzHlvlnk531EvaZ1aHnmkd4E3o0XxyESjUYrbm8xxpgEoAfwPZAJDAG2AP+01mbt7fkjI1R8UvtAc3Y6NGcnVNvmb/wgzNkZBCdozk7uz6HUN6QrLXJrbT4wtcSmW92IIyIi+mWniEjoqZCLiIScCrmISMipkIuIhJwKuYhIyKmQi4iEnAq5iEjIqZCLiIScCrmISMipkIuIhJwKuYhIyFWqqd4qyo9Jl/mdArUDMFjSuqN8HbuMuj/6O2gXwMNvbPU1/rA+tXyNHxT9PvF4nrzdZD/bx9f4jjdKvUctchGRkFMhFxEJORVyEZGQUyEXEQk5FXIRkZBTIRcRCTkVchGRkFMhFxEJORVyEZGQUyEXEQk5FXIRkZBTIRcRCbnKNWhWBHo+A/VbQUEOTB0EW5dXrRziEuCCMVCzMSQkw7yHYek07+I3aQl9boFHBhVvu/A2+GUVzCt9zJ+K5+PrcHByhE/OSKfnx7+SEhfhseNTKYhCTiEM+r8sNuR4NBiZ3+8Hv98Lv27l0IevZP2Nj1Pz3bHE79gMQMLmX8hpcgybBo50Je7X63by6NxfeLVfU5Ztymb4zLVEgRb1Uhh+ZgPi4yKM+Wwj05dsIy0pjkHtDqbzEekHFNP1FrkxxrMh7Fr0goQUGNse3h8G3UZ5FTk4ObTqBzu3wEtd4JXzoOfj3sU+qz/0HwGJsZEb02rBLU/B8ad7l0MRv16HhAg8dUI1dhU4xfrRVqkM/WoX3T/M5O21udxqUrxJBP/PRV/jF+RTZ+I/iCY5J+OmgSNZf8tTbBz8IIWpaWz9042uhH1hwUb+OuMncvILAXjsw/UMPf0QXru0Gdl5hcxetgO7MZtpi7fxn8uaMfbCJjz58Xp25RUeUFxXCrkxppkxZqYx5kcg1xjzmTFmgjHmEDfiFWnUEZbNdJbXLoAGrd2MFswcvpsMH9xbvF5Y4F3sjWvg6VuL11NS4e3n4NPp3uVQxK/X4eHjUnlhZQ4/ZztvzMs/z+Kb7c6LkBCJkF3g3dDAfp+Lfsav9eZT/NqxFwUH1f3d9oOmjeXXTn3+sL2iNKqZxOgLGv+2PrpXI9ocVp3cgkI2ZuVTp1oCyzdl07ZRdZIT4khOiKNxrWTshuwDiutWi/xp4CZrbWPgVOC/wCjgJZfiAZCcDjnbi9ejBRCJdzNi8HLIzYLcTEhKg4smwgcjvIv9vw+gIL94fdM6WPmtd/FL8uN1uLRxEhtzory/vvgg/JLtFO6Ta8dzTbNkRv+Q424SJfh9LvoVv/qn71KYVpPso9v9bnvcr1tJsQvJPPls12J3NweREF/cCREfF2Ht9lzOeekHtu4qoEmdZMzBKSxck0VmTgFbd+Xz5dqdwWyRAwdZa78HsNZ+BnSw1v4PcHWU/JwdkFSjeD0S55w8XgpCDukNYeAs+HoCfDPJ29hB4cfrcEXjJLrUS+C909I47qB4XmpdnfrJEfo0TOTJE6txwSeZbMr1rkXu97noV/y0T6eTsnQh9f95A0k/LaPuK38jbvtmqn05h6w2XSHO29ZdxkFJzBpsuPj42jw8+2ea1U3hkhPrcNXrq/j77J9pdWgqtaodWE5uFfIVxpjnjDHnG2NeAL4yxvQGslyKB8Ca+dC8h7Oc0Q7WL3IzWjBzqF4P+k+HWXfDF+O9jR0kfrwOXT/MpNuHmXT/MJNvthdw5cIsOtdL5JpmyXSfl8mqrANrde0rv89Fv+KvH/o064c8xfpbniK34RFsuvyvFB5Uh9SlC8k++mRvkoi5ZvIqVm1xPoVVT4ojjghbduazdVcBEy9txt1nNuDnX/NoXvfAvjtx66qVAcBVQDfgc2As0Aa4yKV4ACyZAk27wsD5QATeHuBmtGDmcPodkFITOt3p/IHzpWf+gXXBhY7frwNAPDDq+FTW7CzktVOqA/DRxnz+tsSbF8PvY+B3/N0lrF9NXt0GnsYc3O5ghr37E4nxEVIT4vjb2RnUSo3np225/Gn8MhLjI/yl86HExx3YNSGRaNTfeRn3ZGSE4CXlsXzN2UmjrzVnp+bsdDz0342+xs9efa2v8QEY+Eapbwj9IEhEJORUyEVEQk6FXEQk5FTIRURCToVcRCTkVMhFREJOhVxEJORUyEVEQk6FXEQk5FTIRURCToVcRCTkVMhFREIukINmiYhI+alFLiIScirkIiIhp0IuIhJyKuQiIiGnQi4iEnIq5CIiIefW5Mu+MMbEAc8ArYAcYJC1dpkPebQD/m6t7eRD7EScya4PB5KBv1lrp3qcQzzwAmCAAmCAtXa5lznE8qgH/A/oaq1d6kP8L4HtsdWV1lrPpx82xtwJnAckAc9Ya1/yOH5/oH9sNQU4HjjEWrvNo/iJwHic90MBcJXX54IxJhl4GWgK7ACut9b+UJExKluLvBeQYq09BRgGjPI6AWPMX4AXcU5aP1wKbLbWngqcDTzlQw7nAlhrOwD3AI95nUDsDfw8sMvr2LH4KQDW2k6xPz+KeCegPdABOB04zOscrLXjio4Bzn+qN3lVxGN6AAnW2vbAfcADHsYuchWQaa09GbgRF96Tla2QdwRmAlhrPwNa+5DDcqC3D3GLvA4ML7Ge73UC1tq3gMGx1cbAeq9zAB4FngPW+RAbnE+F1Ywxs4wxs40xJ/uQQ3dgETAFeAeY5kMOABhjWgPHWGvHeBz6eyAh9mk9HcjzOD7A0cAMAGutBY6q6ACVrZCnU/xRFqDAGONp95G1djL+nCxF8TOttb8aY2oAbwB/9SmPfGPMeGB0LA/PxD7Ob7TWvudl3N3sxPnPpDtwDfBvr89FoC5OY6ZviRwiHudQ5C5gpA9xM3G6VZbidPc96UMOXwHnGGMisf/QM2LdjxWmshXyHUCNEutx1lrPW6R+M8YcBswBXrXWTvArD2vtFcCRwAvGmOoehh4IdDXGzMXpk33FGHOIh/HBaQn+y1obtdZ+D2wGDvU4h83Ae9ba3FhLMBs42OMcMMbUBFpYa+d4HRsYgnMMjsT5lDS+qNvLQ2NxatMcnG7H/1lrCyoyQGUr5PNx+sSI/c+3yN90vGeMqQ/MAu6w1o71KYfLYl+ygdMyLcT5oskT1trTrLWnx/plvwIut9b+4lX8mIHEvqMxxjTA+bT4s8c5fAycFWsJNgCq4xR3r50GvO9DXICtFH9K3wIkAhXaGi6HNsDHsfNxCrCiogNUqqtWcA5SV2PMJ0AE8PwLpgC4C6gFDDfGFPWVn22t9fJLvzeBl40xH+K8cW6x1mZ7GD8IXgLGGWM+BqLAQK8/HVprpxljTgM+x2m0XV/RLcFyMrhQvMrpcWCsMeYjnCt37rLWZnmcww/A/caY24BtwJUVHUCjH4qIhFxl61oREalyVMhFREJOhVxEJORUyEVEQk6FXEQk5FTIpVIwxhxkjJniwn4PN8as2stj7jXG3FuR+xTZFyrkUlnUAk7wOwkRP1S2HwRJ1fUk0CDWKh+CM3jaJpzRD/8NdLLW9geI/XT/XmvtXGPMMODPOL/2ew/nF7F7/HGFMaYlztgxaUA94CFr7XOxu9saYxbE7htjrX0i9pw/7L+C/90iapFLpXETsM5ae0Fs3QCXWmu7lvYEY8xZwEk4P6E+AcgALikjxiCc8d3bAJ2BR0rcdyhwBnAKcIMx5vj92L/IflGLXCqrDdbaVXt5zJlAO5xxsgFSgdVlPP5WnLFL7gSOxWl9F3mt6Kffxph3KB7/e0/7/7j8/wyRvVMhl8qq5NgyUZyxd4okxm7jgX9aax+D30bpK2s8lP/gDML0DvAacHGJ+0o+Lw5nKOPS9l93X/8xImVR14pUFvmU3jDZBBwVGwWwCXBcbPts4DJjTFpsrPC3gD5lxOgK3GOtfRtn9qWiae0A+hhjko0xtYBzcIYs3df9i+wXFXKpLNYDq40xexrz+n1gDWCBJ4h1bVhr3wEmAwuAb3GGvB1fRox7gY+NMYuBU4FVQJPYfT/iDKP8MfCgtXbJfuxfZL9o9EMRkZBTi1xEJORUyEVEQk6FXEQk5FTIRURCToVcRCTkVMhFREJOhVxEJORUyEVEQu7/AZyFrO34ycpcAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(6, 6))\n",
    "draw_confusionmatix(yy, yyk)\n",
    "plt.title('K-Means',fontsize=20)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### SpectralClustering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-03-04T07:15:04.696143Z",
     "start_time": "2019-03-04T07:15:03.910551Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 482 ms\n",
      "accuracy_score: 0.7412353923205343\n"
     ]
    }
   ],
   "source": [
    "# SpectralClustering\n",
    "sc = SpectralClustering(n_clusters = 10, n_init=100,assign_labels='discretize')\n",
    "%time yys = sc.fit_predict(xx)\n",
    "yys_t = predict_transfer(yy, yys)\n",
    "acc = accuracy_score(yy,yys_t)\n",
    "print('accuracy_score:', acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-03-04T07:15:07.373885Z",
     "start_time": "2019-03-04T07:15:06.725699Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "figure = plt.subplots(1, 2, figsize=(12,6))\n",
    "plt.subplot(1,2,1)\n",
    "draw_confusionmatix(yy, yys)\n",
    "plt.title('SpectralClustering',fontsize=20)\n",
    "plt.subplot(1,2,2)\n",
    "draw_confusionmatix(yy, yys_t)\n",
    "plt.title('SpectralClustering trans',fontsize=20)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### AgglomerativeClustering(Hierarchical Clustering)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-01-15T13:22:25.919980Z",
     "start_time": "2019-01-15T13:22:25.710097Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 173 ms\n",
      "accuracy_score: 0.8352810239287701\n"
     ]
    }
   ],
   "source": [
    "# AgglomerativeClustering(Hierarchical Clustering)\n",
    "hc = AgglomerativeClustering(n_clusters=10)\n",
    "%time yyh = hc.fit_predict(xx)\n",
    "yyh = predict_transfer(yy, yyh)\n",
    "acc = accuracy_score(yy,yyh)\n",
    "print('accuracy_score:', acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-01-15T13:22:26.771277Z",
     "start_time": "2019-01-15T13:22:26.509518Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(6, 6))\n",
    "draw_confusionmatix(yy, yyh)\n",
    "plt.title('AgglomerativeClustering',fontsize=20)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Birch\n",
    "在scikit-learn中，BIRCH类实现了原理篇里讲到的基于特征树CF Tree的聚类。因此要使用BIRCH来聚类，关键是对CF Tree结构参数的处理。\n",
    "\n",
    "在CF Tree中，几个关键的参数为内部节点的最大CF数B， 叶子节点的最大CF数L， 叶节点每个CF的最大样本半径阈值T。这三个参数定了，CF Tree的结构也基本确定了，最后的聚类效果也基本确定。可以说BIRCH的调参就是调试B,L和T。\n",
    "\n",
    "至于类别数K，此时反而是可选的，不输入K，则BIRCH会对CF Tree里各叶子节点CF中样本的情况自己决定类别数K值，如果输入K值，则BIRCH会CF Tree里各叶子节点CF进行合并，直到类别数为K。\n",
    "\n",
    "在scikit-learn中，BIRCH类的重要参数不多，下面一并讲解。\n",
    "\n",
    "1. **threshold**:即叶节点每个CF的最大样本半径阈值T，它决定了每个CF里所有样本形成的超球体的半径阈值。一般来说threshold越小，则CF Tree的建立阶段的规模会越大，即BIRCH算法第一阶段所花的时间和内存会越多。但是选择多大以达到聚类效果则需要通过调参决定。默认值是0.5.如果样本的方差较大，则一般需要增大这个默认值。\n",
    "\n",
    "2. **branching_factor**：即CF Tree内部节点的最大CF数B，以及叶子节点的最大CF数L。这里scikit-learn对这两个参数进行了统一取值。也就是说，branching_factor决定了CF Tree里所有节点的最大CF数。默认是50。如果样本量非常大，比如大于10万，则一般需要增大这个默认值。选择多大的branching_factor以达到聚类效果则需要通过和threshold一起调参决定。\n",
    "\n",
    "3. **n_clusters**：即类别数K，在BIRCH算法是可选的，如果类别数非常多，我们也没有先验知识，则一般输入None，此时BIRCH算法第4阶段不会运行。但是如果我们有类别的先验知识，则推荐输入这个可选的类别值。默认是3，即最终聚为3类。\n",
    "\n",
    "4. **compute_labels**：布尔值，表示是否标示类别输出，默认是True。一般使用默认值挺好，这样可以看到聚类效果。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "D:\\Anaconda\\lib\\site-packages\\sklearn\\cluster\\birch.py:77: FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated; use `arr[tuple(seq)]` instead of `arr[seq]`. In the future this will be interpreted as an array index, `arr[np.array(seq)]`, which will result either in an error or a different result.\n",
      "  node1_dist, node2_dist = dist[[farthest_idx]]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wall time: 234 ms\n",
      "accuracy_score: 0.19421257651641624\n"
     ]
    }
   ],
   "source": [
    "# Birch\n",
    "bir = Birch(n_clusters = 10)\n",
    "%time yyb = bir.fit_predict(xx)\n",
    "yyb_t = predict_transfer(yy, yyb)\n",
    "acc = accuracy_score(yy,yyb)\n",
    "print('accuracy_score:', acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x432 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(6, 6))\n",
    "draw_confusionmatix(yy, yyb_t)\n",
    "plt.title('Birch',fontsize=20)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "figure = plt.figure(figsize=(10, 10))\n",
    "ax1 = figure.add_subplot(2, 2, 1, aspect = 'equal')\n",
    "sns.heatmap(confusion_matrix(yy, yyk).T, square=True, annot=True, fmt='d', cbar=False,cmap = 'rainbow',\n",
    "            yticklabels=dx.target_names, xticklabels=dx.target_names)\n",
    "plt.xlabel('true label')\n",
    "plt.ylabel('predicted label')\n",
    "\n",
    "ax2 = figure.add_subplot(2, 2, 2, aspect = 'equal')\n",
    "sns.heatmap(confusion_matrix(yy, yyh).T, square=True, annot=True, fmt='d', cbar=False,cmap = 'rainbow',\n",
    "            yticklabels=dx.target_names, xticklabels=dx.target_names)\n",
    "plt.xlabel('true label')\n",
    "plt.ylabel('predicted label')\n",
    "\n",
    "ax3 = figure.add_subplot(2, 2, 3, aspect = 'equal')\n",
    "sns.heatmap(confusion_matrix(yy, yys_t).T, square=True, annot=True, fmt='d', cbar=False,cmap = 'rainbow',\n",
    "            yticklabels=dx.target_names, xticklabels=dx.target_names)\n",
    "plt.xlabel('true label')\n",
    "plt.ylabel('predicted label')\n",
    "\n",
    "ax4 = figure.add_subplot(2, 2, 4, aspect = 'equal')\n",
    "sns.heatmap(confusion_matrix(yy, yyb_t).T, square=True, annot=True, fmt='d', cbar=False,cmap = 'rainbow',\n",
    "            yticklabels=dx.target_names, xticklabels=dx.target_names)\n",
    "plt.xlabel('true label')\n",
    "plt.ylabel('predicted label')\n",
    "\n",
    "ax1.set_title(\"K-means\")\n",
    "ax2.set_title(\"Agglomerative Clustering\")\n",
    "ax3.set_title(\"Spectrul Clustering\")\n",
    "ax4.set_title(\"Birch\")\n",
    "\n",
    "\n",
    "figure.suptitle('Clustering')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### DBSCAN\n",
    "\n",
    "根据密度聚类，不需指定聚类中心数"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-01-15T11:56:18.030492Z",
     "start_time": "2019-01-15T11:56:17.524363Z"
    }
   },
   "outputs": [],
   "source": [
    "# DBSCAN\n",
    "%time yyd = DBSCAN(eps = 1.2, min_samples=2).fit_predict(xx)\n",
    "\n",
    "yyd = predict_transfer(yy,yyd)\n",
    "acc = accuracy_score(yy,yyd)\n",
    "print('accuracy_score:', acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-01-15T13:09:44.342898Z",
     "start_time": "2019-01-15T13:09:44.085530Z"
    }
   },
   "outputs": [],
   "source": [
    "plt.figure(figsize=(6, 6))\n",
    "draw_confusionmatix(yy, yyd)\n",
    "plt.title('DBSCAN',fontsize=20)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### MeanShift\n",
    "meanshift不用指定聚类中心数。\n",
    "\n",
    "meanshift聚类步骤如下：\n",
    "1. 随机生成center，把距离center bandwith内的点归为该簇，这些点归为该簇的计数加1\n",
    "2. 计算center到bandwith范围内所有点的距离向量求和，得到向量shift，然后center移动shift，新的点归为该簇的计数加1\n",
    "3. 重复上述步骤直至shift很小\n",
    "4. 产生下一个center，重复以上步骤，根据每个点归为不同簇的次数划分类别"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-01-15T12:00:33.112148Z",
     "start_time": "2019-01-15T12:00:29.372396Z"
    },
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# MeanShift\n",
    "ms = MeanShift(bandwidth = 1.1)\n",
    "%time yym = ms.fit_predict(xx)\n",
    "yym = predict_transfer(yy, yym)\n",
    "acc = accuracy_score(yy,yym)\n",
    "print('accuracy_score:', acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2019-01-15T12:00:37.089606Z",
     "start_time": "2019-01-15T12:00:36.844202Z"
    }
   },
   "outputs": [],
   "source": [
    "draw_confusionmatix(yy, yym)\n",
    "plt.title('MeanShift')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {
    "height": "calc(100% - 180px)",
    "left": "10px",
    "top": "150px",
    "width": "191px"
   },
   "toc_section_display": true,
   "toc_window_display": false
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}