train_rnn_models.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The autoreload extension is already loaded. To reload it, use:\n",
      "  %reload_ext autoreload\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%aimport data_creater\n",
    "%aimport model\n",
    "%autoreload 1\n",
    "\n",
    "from data_creater import *\n",
    "from model import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['MMM', 'AXP', 'AAPL', 'BA', 'CAT', 'CVX', 'CSCO', 'KO', 'DWDP', 'XOM', 'WBA', 'GS', 'HD', 'IBM', 'INTC', 'JNJ', 'JPM', 'MCD', 'MRK', 'MSFT', 'NKE', 'PFE', 'PG', 'TRV', 'UNH', 'UTX', 'VZ', 'V', 'WMT', 'DIS', 'GOOG', 'FB']\n"
     ]
    }
   ],
   "source": [
    "stocks = companies()\n",
    "symbols = stocks['Symbol'].values.tolist()\n",
    "print(symbols)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "window_sizes = [5,7,10]\n",
    "dropouts =  [0.25,0.4]\n",
    "learn_rates = [0.01,0.001]\n",
    "epochs = [100,200]\n",
    "batch_size = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** Best Model Selection for MMM ***\n",
      "============================================================\n",
      "\n",
      "Window size: 5\n",
      "------------------------------------------------------------\n",
      "WARNING:tensorflow:From C:\\Users\\Ashutosh\\Anaconda3\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:1188: calling reduce_sum (from tensorflow.python.ops.math_ops) with keep_dims is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "keep_dims is deprecated, use keepdims instead\n",
      "WARNING:tensorflow:From C:\\Users\\Ashutosh\\Anaconda3\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py:1290: calling reduce_mean (from tensorflow.python.ops.math_ops) with keep_dims is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "keep_dims is deprecated, use keepdims instead\n",
      "1    > Learn rate: 0.0100 Dropout: 0.25 Epoch: 100 Training error: 0.0047 Testing error: 0.0060\n",
      "2    > Learn rate: 0.0100 Dropout: 0.25 Epoch: 200 Training error: 0.0037 Testing error: 0.0070\n",
      "3    > Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0068 Testing error: 0.0061\n",
      "4    > Learn rate: 0.0100 Dropout: 0.40 Epoch: 200 Training error: 0.0043 Testing error: 0.0048\n",
      "5    > Learn rate: 0.0010 Dropout: 0.25 Epoch: 100 Training error: 0.0046 Testing error: 0.0043\n",
      "6    > Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0032 Testing error: 0.0038\n",
      "7    > Learn rate: 0.0010 Dropout: 0.40 Epoch: 100 Training error: 0.0039 Testing error: 0.0040\n",
      "8    > Learn rate: 0.0010 Dropout: 0.40 Epoch: 200 Training error: 0.0035 Testing error: 0.0037\n",
      "\n",
      "Window size: 7\n",
      "------------------------------------------------------------\n",
      "9    > Learn rate: 0.0100 Dropout: 0.25 Epoch: 100 Training error: 0.0032 Testing error: 0.0036\n",
      "10   > Learn rate: 0.0100 Dropout: 0.25 Epoch: 200 Training error: 0.0021 Testing error: 0.0038\n",
      "11   > Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0063 Testing error: 0.0072\n",
      "12   > Learn rate: 0.0100 Dropout: 0.40 Epoch: 200 Training error: 0.0036 Testing error: 0.0041\n",
      "13   > Learn rate: 0.0010 Dropout: 0.25 Epoch: 100 Training error: 0.0039 Testing error: 0.0041\n",
      "14   > Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0031 Testing error: 0.0036\n",
      "15   > Learn rate: 0.0010 Dropout: 0.40 Epoch: 100 Training error: 0.0040 Testing error: 0.0042\n",
      "16   > Learn rate: 0.0010 Dropout: 0.40 Epoch: 200 Training error: 0.0042 Testing error: 0.0038\n",
      "\n",
      "Window size: 10\n",
      "------------------------------------------------------------\n",
      "17   > Learn rate: 0.0100 Dropout: 0.25 Epoch: 100 Training error: 0.0036 Testing error: 0.0041\n",
      "18   > Learn rate: 0.0100 Dropout: 0.25 Epoch: 200 Training error: 0.0021 Testing error: 0.0045\n",
      "19   > Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0035 Testing error: 0.0044\n",
      "20   > Learn rate: 0.0100 Dropout: 0.40 Epoch: 200 Training error: 0.0029 Testing error: 0.0038\n",
      "21   > Learn rate: 0.0010 Dropout: 0.25 Epoch: 100 Training error: 0.0032 Testing error: 0.0041\n",
      "22   > Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0028 Testing error: 0.0035\n",
      "23   > Learn rate: 0.0010 Dropout: 0.40 Epoch: 100 Training error: 0.0032 Testing error: 0.0038\n",
      "24   > Learn rate: 0.0010 Dropout: 0.40 Epoch: 200 Training error: 0.0031 Testing error: 0.0038\n",
      "\n",
      "Model selection summary for MMM with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0028 Testing error: 0.0035\n",
      "\n",
      "Results : \n",
      "------------------------------------------------------------\n",
      "<keras.models.Sequential object at 0x00000288E642D6D8>\n",
      "{'test_error': 0.0035, 'learn_rate': 0.001, 'dropout': 0.25, 'ticker': 'MMM', 'epoch': 200, 'window_size': 10, 'train_error': 0.0028}\n",
      "Saved trained model for MMM\n"
     ]
    }
   ],
   "source": [
    "result = model_selector(symbols[0], window_sizes, learn_rates, dropouts, epochs, batch_size,verbose=1)\n",
    "\n",
    "print(\"\\nResults : \")\n",
    "print(\"-\"*60)\n",
    "print(result[0])\n",
    "\n",
    "print(result[1])\n",
    "ModelLoader.save(result[1]['ticker'],result[0],result[1])\n",
    "print(\"Saved trained model for {}\".format(result[1]['ticker']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Model selection summary for AXP with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0035 Testing error: 0.0053\n",
      " ==> Saved trained model for AXP\n",
      "\n",
      "Model selection summary for AAPL with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0031 Testing error: 0.0109\n",
      " ==> Saved trained model for AAPL\n",
      "\n",
      "Model selection summary for BA with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0059 Testing error: 0.0031\n",
      " ==> Saved trained model for BA\n",
      "\n",
      "Model selection summary for CAT with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0030 Testing error: 0.0055\n",
      " ==> Saved trained model for CAT\n",
      "\n",
      "Model selection summary for CVX with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.40 Epoch: 100 Training error: 0.0070 Testing error: 0.0106\n",
      " ==> Saved trained model for CVX\n",
      "\n",
      "Model selection summary for CSCO with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.25 Epoch: 100 Training error: 0.0037 Testing error: 0.0079\n",
      " ==> Saved trained model for CSCO\n",
      "\n",
      "Model selection summary for KO with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.25 Epoch: 100 Training error: 0.0037 Testing error: 0.0058\n",
      " ==> Saved trained model for KO\n",
      "\n",
      "Model selection summary for DWDP with window size of 5:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0051 Testing error: 0.0058\n",
      " ==> Saved trained model for DWDP\n",
      "\n",
      "Model selection summary for XOM with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 100 Training error: 0.0103 Testing error: 0.0130\n",
      " ==> Saved trained model for XOM\n",
      "\n",
      "Model selection summary for WBA with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.40 Epoch: 100 Training error: 0.0058 Testing error: 0.0100\n",
      " ==> Saved trained model for WBA\n",
      "\n",
      "Model selection summary for GS with window size of 5:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0093 Testing error: 0.0071\n",
      " ==> Saved trained model for GS\n",
      "\n",
      "Model selection summary for HD with window size of 5:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0018 Testing error: 0.0040\n",
      " ==> Saved trained model for HD\n",
      "\n",
      "Model selection summary for IBM with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.25 Epoch: 200 Training error: 0.0058 Testing error: 0.0097\n",
      " ==> Saved trained model for IBM\n",
      "\n",
      "Model selection summary for INTC with window size of 5:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.25 Epoch: 200 Training error: 0.0033 Testing error: 0.0080\n",
      " ==> Saved trained model for INTC\n",
      "\n",
      "Model selection summary for JNJ with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.40 Epoch: 100 Training error: 0.0052 Testing error: 0.0045\n",
      " ==> Saved trained model for JNJ\n",
      "\n",
      "Model selection summary for JPM with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.25 Epoch: 100 Training error: 0.0026 Testing error: 0.0042\n",
      " ==> Saved trained model for JPM\n",
      "\n",
      "Model selection summary for MCD with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0023 Testing error: 0.0041\n",
      " ==> Saved trained model for MCD\n",
      "\n",
      "Model selection summary for MRK with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0036 Testing error: 0.0449\n",
      " ==> Saved trained model for MRK\n",
      "\n",
      "Model selection summary for MSFT with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 200 Training error: 0.0025 Testing error: 0.0062\n",
      " ==> Saved trained model for MSFT\n",
      "\n",
      "Model selection summary for NKE with window size of 5:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0041 Testing error: 0.0125\n",
      " ==> Saved trained model for NKE\n",
      "\n",
      "Model selection summary for PFE with window size of 5:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 200 Training error: 0.0018 Testing error: 0.0497\n",
      " ==> Saved trained model for PFE\n",
      "\n",
      "Model selection summary for PG with window size of 5:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0046 Testing error: 0.0100\n",
      " ==> Saved trained model for PG\n",
      "\n",
      "Model selection summary for TRV with window size of 5:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0041 Testing error: 0.0047\n",
      " ==> Saved trained model for TRV\n",
      "\n",
      "Model selection summary for UNH with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.25 Epoch: 200 Training error: 0.0028 Testing error: 0.0035\n",
      " ==> Saved trained model for UNH\n",
      "\n",
      "Model selection summary for UTX with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0038 Testing error: 0.0102\n",
      " ==> Saved trained model for UTX\n",
      "\n",
      "Model selection summary for VZ with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.25 Epoch: 100 Training error: 0.0033 Testing error: 0.0100\n",
      " ==> Saved trained model for VZ\n",
      "\n",
      "Model selection summary for V with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0024 Testing error: 0.0062\n",
      " ==> Saved trained model for V\n",
      "\n",
      "Model selection summary for WMT with window size of 7:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0028 Testing error: 0.0033\n",
      " ==> Saved trained model for WMT\n",
      "\n",
      "Model selection summary for DIS with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0089 Testing error: 0.0182\n",
      " ==> Saved trained model for DIS\n",
      "\n",
      "Model selection summary for GOOG with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0100 Dropout: 0.40 Epoch: 100 Training error: 0.0028 Testing error: 0.0099\n",
      " ==> Saved trained model for GOOG\n",
      "\n",
      "Model selection summary for FB with window size of 10:\n",
      "------------------------------------------------------------\n",
      " ==> Learn rate: 0.0010 Dropout: 0.25 Epoch: 200 Training error: 0.0022 Testing error: 0.0106\n",
      " ==> Saved trained model for FB\n"
     ]
    }
   ],
   "source": [
    "from keras import backend as K\n",
    "import tensorflow as tf\n",
    "\n",
    "for ticker in symbols[1:]:\n",
    "    #release memory\n",
    "    K.clear_session()\n",
    "    tf.reset_default_graph()\n",
    "    \n",
    "    result = model_selector(ticker, window_sizes, learn_rates, dropouts, epochs, batch_size,verbose=2)\n",
    "    \n",
    "    #save trained model\n",
    "    ModelLoader.save(result[1]['ticker'],result[0],result[1])\n",
    "    print(\" ==> Saved trained model for {}\".format(result[1]['ticker']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.5.5"
  },
  "widgets": {
   "state": {},
   "version": "1.1.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}