transformers_time_series.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from sqlalchemy import create_engine\n",
    "import numpy as np\n",
    "import json\n",
    "import torch\n",
    "import pickle\n",
    "import configparser\n",
    "from ast import literal_eval\n",
    "from hiko.dask import daskify\n",
    "from sklearn.metrics import accuracy_score, roc_auc_score\n",
    "from torch import nn\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "sns.set_style('white')\n",
    "\n",
    "# Read in config\n",
    "config = configparser.ConfigParser()\n",
    "config.read('../ml_models_for_airflow/dbs3_config.ini')\n",
    "\n",
    "db_engine = create_engine(config['AIRFLOW']['postgres_conn'])\n",
    "\n",
    "pairs_mapping = literal_eval(config['MODEL']['pairs_mapping'])\n",
    "pairs = tuple(pairs_mapping.values())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[I 2020-05-14 10:02:25,145] Finished trial#0 with value: 0.6128246753246753\n",
    "with parameters: {'seq_length': 52, 'num_attention_heads': 6, 'num_transformer_blocks': 9,\n",
    "                  'num_out_channels_emb': 191, 'dropout': 0.420575643690622, 'batch_size': 188, 'num_epochs': 125, 'optimizer': 'SGD', 'learning_rate': 7.245336658547346e-05}.\n",
    "Best is trial#0 with value: 0.6128246753246753."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "SEQ_LENGHT = 25\n",
    "NUM_ATTENTION_HEADS = 1\n",
    "NUM_TRANSFORMER_BLOCKS = 1\n",
    "NUM_OUT_CHANNELS_EMB = 2000\n",
    "DROPOUT = 0.420575643690622\n",
    "BATCH_SIZE = 188\n",
    "NUM_EPOCHS = 200\n",
    "OPTIMIZER = 'SGD'\n",
    "LEARNING_RATE = 7.245336658547346e-05\n",
    "GRAD_CLIPPING_VAL = 1\n",
    "INIT_LR = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from training_data_loader import Dataset\n",
    "from torch.utils import data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformer import Transformer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "full_data_set = Dataset(config_location='../ml_models_for_airflow/dbs3_config.ini',\n",
    "                      pairs=pairs,\n",
    "                      seq_lenght=SEQ_LENGHT,\n",
    "                      num_features=2,\n",
    "                      local_path='book_data_raw.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3889"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(full_data_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_set_size = int(len(full_data_set)*0.85)\n",
    "test_set_size = len(full_data_set) - train_set_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainset, valset = data.random_split(full_data_set,\n",
    "                                     [train_set_size, test_set_size]\n",
    "                                    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_generator = data.DataLoader(trainset, batch_size=BATCH_SIZE, shuffle=True, num_workers=4)\n",
    "test_generator = data.DataLoader(valset, batch_size=len(valset), shuffle=True, num_workers=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "transformer = Transformer(emb=SEQ_LENGHT,\n",
    "                          heads=NUM_ATTENTION_HEADS,\n",
    "                          depth=NUM_TRANSFORMER_BLOCKS,\n",
    "                          num_features=2,\n",
    "                          num_out_channels_emb=NUM_OUT_CHANNELS_EMB,\n",
    "                          dropout=DROPOUT\n",
    "                          )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "if OPTIMIZER == 'ADAM': optimizer = torch.optim.Adam(transformer.parameters(), lr=LEARNING_RATE)\n",
    "if OPTIMIZER == 'SGD': optimizer = torch.optim.SGD(transformer.parameters(), lr=LEARNING_RATE)\n",
    "\n",
    "learning_rate_scheduler = torch.optim.lr_scheduler.LambdaLR(optimizer, lambda i: min(i / (INIT_LR / BATCH_SIZE), 1.0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Transformer(\n",
       "  (time_series_features_encoding): Conv1d(2, 2, kernel_size=(1,), stride=(1,), bias=False)\n",
       "  (pos_embedding): Embedding(25, 25)\n",
       "  (transformer_blocks): Sequential(\n",
       "    (0): TransformerBlock(\n",
       "      (attention): SelfAttention(\n",
       "        (tokeys): Linear(in_features=25, out_features=25, bias=False)\n",
       "        (toqueries): Linear(in_features=25, out_features=25, bias=False)\n",
       "        (tovalues): Linear(in_features=25, out_features=25, bias=False)\n",
       "        (unifyheads): Linear(in_features=25, out_features=25, bias=True)\n",
       "      )\n",
       "      (norm1): LayerNorm((25,), eps=1e-05, elementwise_affine=True)\n",
       "      (norm2): LayerNorm((25,), eps=1e-05, elementwise_affine=True)\n",
       "      (feed_forward_layers): Sequential(\n",
       "        (0): Linear(in_features=25, out_features=100, bias=True)\n",
       "        (1): ReLU()\n",
       "        (2): Linear(in_features=100, out_features=25, bias=True)\n",
       "      )\n",
       "      (dropout): Dropout(p=0.420575643690622, inplace=False)\n",
       "    )\n",
       "  )\n",
       "  (conv1d_for_embeddings): Conv1d(2, 2000, kernel_size=(1,), stride=(1,), bias=False)\n",
       "  (feed_forward): Linear(in_features=25, out_features=1, bias=True)\n",
       "  (dropout): Dropout(p=0.420575643690622, inplace=False)\n",
       ")"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "transformer.apply(Transformer.init_weights)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "criterion = torch.nn.BCELoss()\n",
    "track_epoch_loss = []\n",
    "train_auc = []\n",
    "test_auc = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train auc: 0.4830993740701661  epoch: 0\n",
      "test auc: 0.6034986634986634  epoch: 0\n",
      "train auc: 0.49560759147132  epoch: 10\n",
      "test auc: 0.6014612414612415  epoch: 10\n",
      "train auc: 0.48780383810343714  epoch: 20\n",
      "test auc: 0.5985328185328185  epoch: 20\n",
      "train auc: 0.5025828513945838  epoch: 30\n",
      "test auc: 0.5939055539055539  epoch: 30\n",
      "train auc: 0.4932532825874331  epoch: 40\n",
      "test auc: 0.5922542322542322  epoch: 40\n",
      "train auc: 0.49844614859450076  epoch: 50\n",
      "test auc: 0.5871101871101871  epoch: 50\n",
      "train auc: 0.5039706761573381  epoch: 60\n",
      "test auc: 0.5766973566973567  epoch: 60\n",
      "train auc: 0.525148563017441  epoch: 70\n",
      "test auc: 0.565981585981586  epoch: 70\n",
      "train auc: 0.517604726851538  epoch: 80\n",
      "test auc: 0.5484288684288684  epoch: 80\n",
      "train auc: 0.4996958079409705  epoch: 90\n",
      "test auc: 0.5272943272943273  epoch: 90\n",
      "train auc: 0.5054479388776894  epoch: 100\n",
      "test auc: 0.5207246807246807  epoch: 100\n",
      "train auc: 0.48859431262016284  epoch: 110\n",
      "test auc: 0.5197267597267597  epoch: 110\n",
      "train auc: 0.5119658707152352  epoch: 120\n",
      "test auc: 0.5196614196614197  epoch: 120\n",
      "train auc: 0.5037501362404204  epoch: 130\n",
      "test auc: 0.5188654588654589  epoch: 130\n",
      "train auc: 0.48999760078688603  epoch: 140\n",
      "test auc: 0.5378081378081379  epoch: 140\n",
      "train auc: 0.4987039889328503  epoch: 150\n",
      "test auc: 0.5506623106623106  epoch: 150\n",
      "train auc: 0.4856633338931998  epoch: 160\n",
      "test auc: 0.5404217404217404  epoch: 160\n",
      "train auc: 0.5034753468316476  epoch: 170\n",
      "test auc: 0.495016335016335  epoch: 170\n",
      "train auc: 0.5116915363608397  epoch: 180\n",
      "test auc: 0.45955449955449956  epoch: 180\n",
      "train auc: 0.5063167460715312  epoch: 190\n",
      "test auc: 0.4597386397386397  epoch: 190\n"
     ]
    }
   ],
   "source": [
    "for ep in range(NUM_EPOCHS):\n",
    "    transformer.train()\n",
    "    epoch_loss = 0\n",
    "    temp_train_auc = 0\n",
    "    \n",
    "    for train_x, train_y in train_generator:\n",
    "        \n",
    "        predictions = transformer(train_x)\n",
    "        loss = criterion(predictions, train_y.view(-1, 1))\n",
    "        epoch_loss += loss.item()\n",
    "        temp_train_auc += roc_auc_score(\n",
    "            train_y.numpy(), predictions.detach().numpy())\n",
    "        \n",
    "        optimizer.zero_grad()\n",
    "        loss.backward()\n",
    "        \n",
    "        # clip up\n",
    "        torch.nn.utils.clip_grad_norm_(transformer.parameters(), GRAD_CLIPPING_VAL)\n",
    "\n",
    "        optimizer.step()\n",
    "        learning_rate_scheduler.step()\n",
    "    \n",
    "    train_auc.append(temp_train_auc/len(train_generator))\n",
    "    if ep % 10 == 0: print('train auc:', train_auc[-1], ' epoch:', ep)\n",
    "    \n",
    "    with torch.no_grad():\n",
    "        transformer.eval()\n",
    "        temp_test_auc = 0\n",
    "        for test_x, test_y in test_generator:\n",
    "            predictions = transformer(test_x)\n",
    "            temp_test_auc += roc_auc_score(\n",
    "                test_y.numpy(), predictions.numpy())\n",
    "\n",
    "    test_auc.append(temp_test_auc/len(test_generator))\n",
    "    if ep % 10 == 0: print('test auc:', test_auc[-1], ' epoch:', ep)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12, 8))\n",
    "plt.title('Test & Training AUC')\n",
    "plt.plot(test_auc, label='test AUC', color='r', alpha=0.7)\n",
    "plt.plot(train_auc, label='train AUC', color='b', alpha=0.7)\n",
    "plt.ylabel('AUC Score')\n",
    "plt.xlabel('Epoch')\n",
    "plt.grid()\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "pp37",
   "language": "python",
   "name": "pp37"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}