Capstone-Modelling-Part-4.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#importing required packages\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import random "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#importing the datasets\n",
    "train_data = pd.read_csv('train.csv')\n",
    "features_data=pd.read_csv('features.csv')\n",
    "stores_data=pd.read_csv('stores.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(421570, 15)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Store</th>\n",
       "      <th>Dept</th>\n",
       "      <th>Date</th>\n",
       "      <th>Weekly_Sales</th>\n",
       "      <th>IsHoliday_x</th>\n",
       "      <th>Temperature</th>\n",
       "      <th>Fuel_Price</th>\n",
       "      <th>MarkDown1</th>\n",
       "      <th>MarkDown2</th>\n",
       "      <th>MarkDown3</th>\n",
       "      <th>MarkDown4</th>\n",
       "      <th>MarkDown5</th>\n",
       "      <th>CPI</th>\n",
       "      <th>Unemployment</th>\n",
       "      <th>IsHoliday_y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-05</td>\n",
       "      <td>24924.50</td>\n",
       "      <td>False</td>\n",
       "      <td>42.31</td>\n",
       "      <td>2.572</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>211.096358</td>\n",
       "      <td>8.106</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-12</td>\n",
       "      <td>46039.49</td>\n",
       "      <td>True</td>\n",
       "      <td>38.51</td>\n",
       "      <td>2.548</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>211.242170</td>\n",
       "      <td>8.106</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-19</td>\n",
       "      <td>41595.55</td>\n",
       "      <td>False</td>\n",
       "      <td>39.93</td>\n",
       "      <td>2.514</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>211.289143</td>\n",
       "      <td>8.106</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-26</td>\n",
       "      <td>19403.54</td>\n",
       "      <td>False</td>\n",
       "      <td>46.63</td>\n",
       "      <td>2.561</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>211.319643</td>\n",
       "      <td>8.106</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-03-05</td>\n",
       "      <td>21827.90</td>\n",
       "      <td>False</td>\n",
       "      <td>46.50</td>\n",
       "      <td>2.625</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>211.350143</td>\n",
       "      <td>8.106</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Store  Dept        Date  Weekly_Sales  IsHoliday_x  Temperature  \\\n",
       "0      1     1  2010-02-05      24924.50        False        42.31   \n",
       "1      1     1  2010-02-12      46039.49         True        38.51   \n",
       "2      1     1  2010-02-19      41595.55        False        39.93   \n",
       "3      1     1  2010-02-26      19403.54        False        46.63   \n",
       "4      1     1  2010-03-05      21827.90        False        46.50   \n",
       "\n",
       "   Fuel_Price  MarkDown1  MarkDown2  MarkDown3  MarkDown4  MarkDown5  \\\n",
       "0       2.572        NaN        NaN        NaN        NaN        NaN   \n",
       "1       2.548        NaN        NaN        NaN        NaN        NaN   \n",
       "2       2.514        NaN        NaN        NaN        NaN        NaN   \n",
       "3       2.561        NaN        NaN        NaN        NaN        NaN   \n",
       "4       2.625        NaN        NaN        NaN        NaN        NaN   \n",
       "\n",
       "          CPI  Unemployment  IsHoliday_y  \n",
       "0  211.096358         8.106        False  \n",
       "1  211.242170         8.106         True  \n",
       "2  211.289143         8.106        False  \n",
       "3  211.319643         8.106        False  \n",
       "4  211.350143         8.106        False  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#merging the three datasets  \n",
    "df=train_data.merge(features_data,how='left',on=['Store','Date'])\n",
    "print(df.shape)\n",
    "df.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(421570, 17)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Store</th>\n",
       "      <th>Dept</th>\n",
       "      <th>Date</th>\n",
       "      <th>Weekly_Sales</th>\n",
       "      <th>IsHoliday_x</th>\n",
       "      <th>Temperature</th>\n",
       "      <th>Fuel_Price</th>\n",
       "      <th>MarkDown1</th>\n",
       "      <th>MarkDown2</th>\n",
       "      <th>MarkDown3</th>\n",
       "      <th>MarkDown4</th>\n",
       "      <th>MarkDown5</th>\n",
       "      <th>CPI</th>\n",
       "      <th>Unemployment</th>\n",
       "      <th>IsHoliday_y</th>\n",
       "      <th>Type</th>\n",
       "      <th>Size</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-05</td>\n",
       "      <td>24924.50</td>\n",
       "      <td>False</td>\n",
       "      <td>42.31</td>\n",
       "      <td>2.572</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>211.096358</td>\n",
       "      <td>8.106</td>\n",
       "      <td>False</td>\n",
       "      <td>A</td>\n",
       "      <td>151315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-12</td>\n",
       "      <td>46039.49</td>\n",
       "      <td>True</td>\n",
       "      <td>38.51</td>\n",
       "      <td>2.548</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>211.242170</td>\n",
       "      <td>8.106</td>\n",
       "      <td>True</td>\n",
       "      <td>A</td>\n",
       "      <td>151315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-19</td>\n",
       "      <td>41595.55</td>\n",
       "      <td>False</td>\n",
       "      <td>39.93</td>\n",
       "      <td>2.514</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>211.289143</td>\n",
       "      <td>8.106</td>\n",
       "      <td>False</td>\n",
       "      <td>A</td>\n",
       "      <td>151315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-26</td>\n",
       "      <td>19403.54</td>\n",
       "      <td>False</td>\n",
       "      <td>46.63</td>\n",
       "      <td>2.561</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>211.319643</td>\n",
       "      <td>8.106</td>\n",
       "      <td>False</td>\n",
       "      <td>A</td>\n",
       "      <td>151315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-03-05</td>\n",
       "      <td>21827.90</td>\n",
       "      <td>False</td>\n",
       "      <td>46.50</td>\n",
       "      <td>2.625</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>211.350143</td>\n",
       "      <td>8.106</td>\n",
       "      <td>False</td>\n",
       "      <td>A</td>\n",
       "      <td>151315</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Store  Dept        Date  Weekly_Sales  IsHoliday_x  Temperature  \\\n",
       "0      1     1  2010-02-05      24924.50        False        42.31   \n",
       "1      1     1  2010-02-12      46039.49         True        38.51   \n",
       "2      1     1  2010-02-19      41595.55        False        39.93   \n",
       "3      1     1  2010-02-26      19403.54        False        46.63   \n",
       "4      1     1  2010-03-05      21827.90        False        46.50   \n",
       "\n",
       "   Fuel_Price  MarkDown1  MarkDown2  MarkDown3  MarkDown4  MarkDown5  \\\n",
       "0       2.572        NaN        NaN        NaN        NaN        NaN   \n",
       "1       2.548        NaN        NaN        NaN        NaN        NaN   \n",
       "2       2.514        NaN        NaN        NaN        NaN        NaN   \n",
       "3       2.561        NaN        NaN        NaN        NaN        NaN   \n",
       "4       2.625        NaN        NaN        NaN        NaN        NaN   \n",
       "\n",
       "          CPI  Unemployment  IsHoliday_y Type    Size  \n",
       "0  211.096358         8.106        False    A  151315  \n",
       "1  211.242170         8.106         True    A  151315  \n",
       "2  211.289143         8.106        False    A  151315  \n",
       "3  211.319643         8.106        False    A  151315  \n",
       "4  211.350143         8.106        False    A  151315  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df=df.merge(stores_data,how='left',on='Store')\n",
    "print(df.shape)\n",
    "df.head(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Store                0\n",
       "Dept                 0\n",
       "Date                 0\n",
       "Weekly_Sales         0\n",
       "IsHoliday_x          0\n",
       "Temperature          0\n",
       "Fuel_Price           0\n",
       "MarkDown1       270889\n",
       "MarkDown2       310322\n",
       "MarkDown3       284479\n",
       "MarkDown4       286603\n",
       "MarkDown5       270138\n",
       "CPI                  0\n",
       "Unemployment         0\n",
       "IsHoliday_y          0\n",
       "Type                 0\n",
       "Size                 0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Check Nans\n",
    "df.isna().sum(axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Fill Nans with zero as Nans are present only in Markdowns where filling with zeroes is okay.\n",
    "df_new=df.fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Since negative values exist in three columns,lets replace with zero\n",
    "df_new.loc[df_new['Weekly_Sales'] < 0.0,'Weekly_Sales'] = 0.0\n",
    "df_new.loc[df_new['MarkDown2'] < 0.0,'MarkDown2'] = 0.0\n",
    "df_new.loc[df_new['MarkDown3'] < 0.0,'MarkDown3'] = 0.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Store</th>\n",
       "      <th>Dept</th>\n",
       "      <th>Date</th>\n",
       "      <th>Weekly_Sales</th>\n",
       "      <th>IsHoliday_x</th>\n",
       "      <th>Temperature</th>\n",
       "      <th>Fuel_Price</th>\n",
       "      <th>MarkDown1</th>\n",
       "      <th>MarkDown2</th>\n",
       "      <th>MarkDown3</th>\n",
       "      <th>MarkDown4</th>\n",
       "      <th>MarkDown5</th>\n",
       "      <th>CPI</th>\n",
       "      <th>Unemployment</th>\n",
       "      <th>Type</th>\n",
       "      <th>Size</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-05</td>\n",
       "      <td>24924.50</td>\n",
       "      <td>False</td>\n",
       "      <td>42.31</td>\n",
       "      <td>2.572</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>211.096358</td>\n",
       "      <td>8.106</td>\n",
       "      <td>A</td>\n",
       "      <td>151315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-12</td>\n",
       "      <td>46039.49</td>\n",
       "      <td>True</td>\n",
       "      <td>38.51</td>\n",
       "      <td>2.548</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>211.242170</td>\n",
       "      <td>8.106</td>\n",
       "      <td>A</td>\n",
       "      <td>151315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-19</td>\n",
       "      <td>41595.55</td>\n",
       "      <td>False</td>\n",
       "      <td>39.93</td>\n",
       "      <td>2.514</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>211.289143</td>\n",
       "      <td>8.106</td>\n",
       "      <td>A</td>\n",
       "      <td>151315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2010-02-26</td>\n",
       "      <td>19403.54</td>\n",
       "      <td>False</td>\n",
       "      <td>46.63</td>\n",
       "      <td>2.561</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>211.319643</td>\n",
       "      <td>8.106</td>\n",
       "      <td>A</td>\n",
       "      <td>151315</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Store  Dept        Date  Weekly_Sales  IsHoliday_x  Temperature  \\\n",
       "0      1     1  2010-02-05      24924.50        False        42.31   \n",
       "1      1     1  2010-02-12      46039.49         True        38.51   \n",
       "2      1     1  2010-02-19      41595.55        False        39.93   \n",
       "3      1     1  2010-02-26      19403.54        False        46.63   \n",
       "\n",
       "   Fuel_Price  MarkDown1  MarkDown2  MarkDown3  MarkDown4  MarkDown5  \\\n",
       "0       2.572        0.0        0.0        0.0        0.0        0.0   \n",
       "1       2.548        0.0        0.0        0.0        0.0        0.0   \n",
       "2       2.514        0.0        0.0        0.0        0.0        0.0   \n",
       "3       2.561        0.0        0.0        0.0        0.0        0.0   \n",
       "\n",
       "          CPI  Unemployment Type    Size  \n",
       "0  211.096358         8.106    A  151315  \n",
       "1  211.242170         8.106    A  151315  \n",
       "2  211.289143         8.106    A  151315  \n",
       "3  211.319643         8.106    A  151315  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_new=df_new.drop(['IsHoliday_y'], axis=1)\n",
    "df_new.head(4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import keras\n",
    "from sklearn.preprocessing import MinMaxScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "train=df_new"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = train[['Store', 'Dept', 'Date', 'Unemployment', 'IsHoliday_x', 'Type', 'Size',\n",
    "       'Temperature', 'Fuel_Price', 'MarkDown1', 'MarkDown2', 'MarkDown3',\n",
    "       'MarkDown4', 'MarkDown5', 'CPI', 'Weekly_Sales']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2010-02-05 2012-10-26\n"
     ]
    }
   ],
   "source": [
    "print(train['Date'].min(), train['Date'].max())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Store             int64\n",
       "Dept              int64\n",
       "Date             object\n",
       "Unemployment    float64\n",
       "IsHoliday_x        bool\n",
       "Type             object\n",
       "Size              int64\n",
       "Temperature     float64\n",
       "Fuel_Price      float64\n",
       "MarkDown1       float64\n",
       "MarkDown2       float64\n",
       "MarkDown3       float64\n",
       "MarkDown4       float64\n",
       "MarkDown5       float64\n",
       "CPI             float64\n",
       "Weekly_Sales    float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Encoding non numerical columns\n",
    "from sklearn import preprocessing\n",
    "cat_col = ['IsHoliday_x','Type']\n",
    "for col in cat_col:\n",
    "    lbl = preprocessing.LabelEncoder()\n",
    "    lbl.fit(train[col].values.astype('str'))\n",
    "    train[col] = lbl.transform(train[col].values.astype('str'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Store             int64\n",
       "Dept              int64\n",
       "Date             object\n",
       "Unemployment    float64\n",
       "IsHoliday_x       int64\n",
       "Type              int64\n",
       "Size              int64\n",
       "Temperature     float64\n",
       "Fuel_Price      float64\n",
       "MarkDown1       float64\n",
       "MarkDown2       float64\n",
       "MarkDown3       float64\n",
       "MarkDown4       float64\n",
       "MarkDown5       float64\n",
       "CPI             float64\n",
       "Weekly_Sales    float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Store                    int64\n",
       "Dept                     int64\n",
       "Date            datetime64[ns]\n",
       "Unemployment           float64\n",
       "IsHoliday_x              int64\n",
       "Type                     int64\n",
       "Size                     int64\n",
       "Temperature            float64\n",
       "Fuel_Price             float64\n",
       "MarkDown1              float64\n",
       "MarkDown2              float64\n",
       "MarkDown3              float64\n",
       "MarkDown4              float64\n",
       "MarkDown5              float64\n",
       "CPI                    float64\n",
       "Weekly_Sales           float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Changing type of Date column\n",
    "train['Date']=pd.to_datetime(train['Date'],format='%Y-%m-%d')\n",
    "train.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "datetime.datetime(2012, 8, 24, 0, 0)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "split_date = pd.datetime(2012,8,24)\n",
    "split_date"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Splitting data to train and dev set\n",
    "train_set = train.loc[train['Date'] <= split_date]\n",
    "dev_set = train.loc[train['Date'] > split_date]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "#  Split dev set into validation and test set\n",
    "split_date_dev = pd.datetime(2012,9,25)\n",
    "val_set = dev_set.loc[dev_set['Date'] <= split_date_dev]\n",
    "test_set = dev_set.loc[dev_set['Date'] > split_date_dev]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Changing index to be date\n",
    "train_set = train_set.set_index('Date')\n",
    "val_set = val_set.set_index('Date')\n",
    "test_set = test_set.set_index('Date')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_set_array = train_set.iloc[:,:].values\n",
    "val_set_array = val_set.iloc[:,:].values\n",
    "test_set_array = test_set.iloc[:,:].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of train, val and test array:\n",
      " (394879, 15) \n",
      " (11854, 15) \n",
      " (14837, 15)\n"
     ]
    }
   ],
   "source": [
    "print(\"Shape of train, val and test array:\\n\",train_set_array.shape,\"\\n\",val_set_array.shape,\"\\n\",test_set_array.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(394879, 15) (11854, 15) (14837, 15)\n"
     ]
    }
   ],
   "source": [
    "# Scaling\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "sc = MinMaxScaler(feature_range=(0,1))\n",
    "train_set_scaled = sc.fit_transform(train_set_array[:,:])\n",
    "val_set_scaled = sc.fit_transform(val_set_array[:,:])\n",
    "test_set_scaled = sc.fit_transform(test_set_array[:,:])\n",
    "\n",
    "print(train_set_scaled.shape, val_set_scaled.shape, test_set_scaled.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(394879, 14) (394879,) (11854, 14) (11854,) (14837, 14) (14837,)\n"
     ]
    }
   ],
   "source": [
    "X_train = []\n",
    "y_train = []\n",
    "X_val = []\n",
    "y_val = []\n",
    "X_test = []\n",
    "y_test = []\n",
    "\n",
    "X_train, y_train = train_set_scaled[:,:-1], train_set_scaled[:,-1]\n",
    "X_val, y_val = val_set_scaled[:,:-1], val_set_scaled[:,-1]\n",
    "X_test, y_test = test_set_scaled[:,:-1], test_set_scaled[:,-1]\n",
    "\n",
    "print(X_train.shape, y_train.shape, X_val.shape, y_val.shape, X_test.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create data structure\n",
    "\n",
    "X_train = X_train.reshape((X_train.shape[0], 1, X_train.shape[1]))\n",
    "X_val = X_val.reshape((X_val.shape[0], 1, X_val.shape[1]))\n",
    "X_test = X_test.reshape((X_test.shape[0], 1, X_test.shape[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(394879, 1, 14) (11854, 1, 14) (14837, 1, 14)\n"
     ]
    }
   ],
   "source": [
    "print(X_train.shape, X_val.shape, X_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import LSTM\n",
    "from keras.layers import Dropout\n",
    "from keras.layers import Flatten"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Initialising RNN\n",
    "regressor = Sequential()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "#adding first  LSTM layer and some dropout regularization\n",
    "regressor.add(LSTM(units = 10, return_sequences = True, activation = 'relu', input_shape = (X_train.shape[1], 14)))\n",
    "regressor.add(Dropout(0.5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "#adding a second LSTM layer and some dropout regularization\n",
    "regressor.add(LSTM(units = 10, return_sequences = True, activation = 'relu'))\n",
    "regressor.add(Dropout(0.5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # adding a third LSTM layer and some dropout regularization\n",
    "regressor.add(LSTM(units = 10, return_sequences = True, activation = 'relu'))\n",
    "regressor.add(Dropout(0.7))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "# adding a fourth LSTM layer and some dropout regularization\n",
    "regressor.add(LSTM(units = 10, return_sequences = False, activation = 'relu'))\n",
    "regressor.add(Dropout(0.5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Adding the output layer\n",
    "#regressor.add(Flatten())\n",
    "regressor.add(Dense(units=1, activation = 'sigmoid'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Compiling the RNN\n",
    "regressor.compile(optimizer='adam', \n",
    "                  loss='mean_squared_error', \n",
    "                  metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 394879 samples, validate on 11854 samples\n",
      "Epoch 1/20\n",
      "394879/394879 [==============================] - 7s 17us/step - loss: 0.0466 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 2/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0043 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 3/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0029 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 4/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0026 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 5/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0024 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 6/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0023 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 7/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0021 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 8/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0021 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 9/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0020 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 10/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0019 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 11/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0018 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 12/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0018 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 13/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0017 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 14/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0017 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 15/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0017 - acc: 0.0032 - val_loss: 0.0200 - val_acc: 0.0028\n",
      "Epoch 16/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0014 - acc: 0.0032 - val_loss: 0.0170 - val_acc: 0.0028\n",
      "Epoch 17/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0011 - acc: 0.0032 - val_loss: 0.0166 - val_acc: 0.0028\n",
      "Epoch 18/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0011 - acc: 0.0032 - val_loss: 0.0164 - val_acc: 0.0028\n",
      "Epoch 19/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0010 - acc: 0.0032 - val_loss: 0.0162 - val_acc: 0.0028\n",
      "Epoch 20/20\n",
      "394879/394879 [==============================] - 4s 11us/step - loss: 0.0010 - acc: 0.0032 - val_loss: 0.0161 - val_acc: 0.0028\n"
     ]
    }
   ],
   "source": [
    "# Fitting the RNN to the training set\n",
    "history = regressor.fit(X_train, \n",
    "              y_train, \n",
    "              epochs = 20, #Can be made 13 from avlidation loss graph\n",
    "              batch_size = 512, \n",
    "              validation_data = (X_val, y_val),\n",
    "              verbose = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_sales = regressor.predict(X_test)\n",
    "predicted_sales_train = regressor.predict(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Reshape X_test for inverse scaling\n",
    "X_test = X_test.reshape((X_test.shape[0], X_test.shape[2]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(14837, 14) (14837, 1)\n"
     ]
    }
   ],
   "source": [
    "print(X_test[:,:].shape, predicted_sales.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Concatenate in the same order-X_test[:,:] followed by predicted_sales\n",
    "predicted_weekly_sales = np.concatenate((X_test[:,:], predicted_sales),axis=1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[6539.69709678],\n",
       "       [6539.69709678],\n",
       "       [6539.69709678],\n",
       "       ...,\n",
       "       [2446.6414043 ],\n",
       "       [2264.81539058],\n",
       "       [2911.29185809]])"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predicted_weekly_sales = sc.inverse_transform(predicted_weekly_sales)\n",
    "predicted_weekly_sales = predicted_weekly_sales[:,14:15]\n",
    "predicted_weekly_sales"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_test = y_test.reshape((y_test.shape[0],1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[18947.81],\n",
       "       [21904.47],\n",
       "       [22764.01],\n",
       "       ...,\n",
       "       [ 1061.02],\n",
       "       [  760.01],\n",
       "       [ 1076.8 ]])"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "observed_weekly_sales = np.concatenate((X_test[:,:], y_test),axis=1)\n",
    "observed_weekly_sales = sc.inverse_transform(observed_weekly_sales)\n",
    "observed_weekly_sales = observed_weekly_sales[:,14:15]\n",
    "observed_weekly_sales"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(14837, 1) (14837, 1)\n"
     ]
    }
   ],
   "source": [
    "print(observed_weekly_sales.shape, predicted_weekly_sales.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Date</th>\n",
       "      <th>Observed Sales</th>\n",
       "      <th>Predicted Sales</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012-09-28</td>\n",
       "      <td>18947.81</td>\n",
       "      <td>6539.697097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2012-10-05</td>\n",
       "      <td>21904.47</td>\n",
       "      <td>6539.697097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2012-10-12</td>\n",
       "      <td>22764.01</td>\n",
       "      <td>6539.697097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012-10-19</td>\n",
       "      <td>24185.27</td>\n",
       "      <td>6293.257222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2012-10-26</td>\n",
       "      <td>27390.81</td>\n",
       "      <td>5727.487281</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Date  Observed Sales  Predicted Sales\n",
       "0 2012-09-28        18947.81      6539.697097\n",
       "1 2012-10-05        21904.47      6539.697097\n",
       "2 2012-10-12        22764.01      6539.697097\n",
       "3 2012-10-19        24185.27      6293.257222\n",
       "4 2012-10-26        27390.81      5727.487281"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "obs_pred_df = pd.DataFrame({'Date': test_set.index,'Observed Sales': observed_weekly_sales.reshape(observed_weekly_sales.shape[0]),'Predicted Sales': predicted_weekly_sales.reshape(predicted_weekly_sales.shape[0])})\n",
    "\n",
    "obs_pred_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Observed Sales</th>\n",
       "      <th>Predicted Sales</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2012-09-28</th>\n",
       "      <td>18947.81</td>\n",
       "      <td>6539.697097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2012-10-05</th>\n",
       "      <td>21904.47</td>\n",
       "      <td>6539.697097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2012-10-12</th>\n",
       "      <td>22764.01</td>\n",
       "      <td>6539.697097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2012-10-19</th>\n",
       "      <td>24185.27</td>\n",
       "      <td>6293.257222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2012-10-26</th>\n",
       "      <td>27390.81</td>\n",
       "      <td>5727.487281</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            Observed Sales  Predicted Sales\n",
       "Date                                       \n",
       "2012-09-28        18947.81      6539.697097\n",
       "2012-10-05        21904.47      6539.697097\n",
       "2012-10-12        22764.01      6539.697097\n",
       "2012-10-19        24185.27      6293.257222\n",
       "2012-10-26        27390.81      5727.487281"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "obs_pred_df.set_index('Date', inplace=True)\n",
    "obs_pred_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Date</th>\n",
       "      <th>Store</th>\n",
       "      <th>Dept</th>\n",
       "      <th>Unemployment</th>\n",
       "      <th>IsHoliday_x</th>\n",
       "      <th>Type</th>\n",
       "      <th>Size</th>\n",
       "      <th>Temperature</th>\n",
       "      <th>Fuel_Price</th>\n",
       "      <th>MarkDown1</th>\n",
       "      <th>MarkDown2</th>\n",
       "      <th>MarkDown3</th>\n",
       "      <th>MarkDown4</th>\n",
       "      <th>MarkDown5</th>\n",
       "      <th>CPI</th>\n",
       "      <th>Weekly_Sales</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012-09-28</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6.908</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>151315</td>\n",
       "      <td>76.08</td>\n",
       "      <td>3.666</td>\n",
       "      <td>3666.27</td>\n",
       "      <td>7.64</td>\n",
       "      <td>1.65</td>\n",
       "      <td>1417.96</td>\n",
       "      <td>4744.28</td>\n",
       "      <td>222.981658</td>\n",
       "      <td>18947.81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2012-10-05</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6.573</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>151315</td>\n",
       "      <td>68.55</td>\n",
       "      <td>3.617</td>\n",
       "      <td>8077.89</td>\n",
       "      <td>0.00</td>\n",
       "      <td>18.22</td>\n",
       "      <td>3617.43</td>\n",
       "      <td>3626.14</td>\n",
       "      <td>223.181477</td>\n",
       "      <td>21904.47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2012-10-12</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6.573</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>151315</td>\n",
       "      <td>62.99</td>\n",
       "      <td>3.601</td>\n",
       "      <td>2086.18</td>\n",
       "      <td>0.00</td>\n",
       "      <td>8.11</td>\n",
       "      <td>602.36</td>\n",
       "      <td>5926.45</td>\n",
       "      <td>223.381296</td>\n",
       "      <td>22764.01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012-10-19</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6.573</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>151315</td>\n",
       "      <td>67.97</td>\n",
       "      <td>3.594</td>\n",
       "      <td>950.33</td>\n",
       "      <td>0.00</td>\n",
       "      <td>4.93</td>\n",
       "      <td>80.25</td>\n",
       "      <td>2312.85</td>\n",
       "      <td>223.425723</td>\n",
       "      <td>24185.27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2012-10-26</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6.573</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>151315</td>\n",
       "      <td>69.16</td>\n",
       "      <td>3.506</td>\n",
       "      <td>2585.85</td>\n",
       "      <td>31.75</td>\n",
       "      <td>6.00</td>\n",
       "      <td>1057.16</td>\n",
       "      <td>1305.01</td>\n",
       "      <td>223.444251</td>\n",
       "      <td>27390.81</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Date  Store  Dept  Unemployment  IsHoliday_x  Type    Size  \\\n",
       "0 2012-09-28      1     1         6.908            0     0  151315   \n",
       "1 2012-10-05      1     1         6.573            0     0  151315   \n",
       "2 2012-10-12      1     1         6.573            0     0  151315   \n",
       "3 2012-10-19      1     1         6.573            0     0  151315   \n",
       "4 2012-10-26      1     1         6.573            0     0  151315   \n",
       "\n",
       "   Temperature  Fuel_Price  MarkDown1  MarkDown2  MarkDown3  MarkDown4  \\\n",
       "0        76.08       3.666    3666.27       7.64       1.65    1417.96   \n",
       "1        68.55       3.617    8077.89       0.00      18.22    3617.43   \n",
       "2        62.99       3.601    2086.18       0.00       8.11     602.36   \n",
       "3        67.97       3.594     950.33       0.00       4.93      80.25   \n",
       "4        69.16       3.506    2585.85      31.75       6.00    1057.16   \n",
       "\n",
       "   MarkDown5         CPI  Weekly_Sales  \n",
       "0    4744.28  222.981658      18947.81  \n",
       "1    3626.14  223.181477      21904.47  \n",
       "2    5926.45  223.381296      22764.01  \n",
       "3    2312.85  223.425723      24185.27  \n",
       "4    1305.01  223.444251      27390.81  "
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_set = test_set.reset_index()\n",
    "test_set.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Date</th>\n",
       "      <th>Weight</th>\n",
       "      <th>Observed_Values</th>\n",
       "      <th>Predicted_Values</th>\n",
       "      <th>Derived</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2012-09-28</td>\n",
       "      <td>1</td>\n",
       "      <td>18947.81</td>\n",
       "      <td>6539.697097</td>\n",
       "      <td>12408.112903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2012-10-05</td>\n",
       "      <td>1</td>\n",
       "      <td>21904.47</td>\n",
       "      <td>6539.697097</td>\n",
       "      <td>15364.772903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2012-10-12</td>\n",
       "      <td>1</td>\n",
       "      <td>22764.01</td>\n",
       "      <td>6539.697097</td>\n",
       "      <td>16224.312903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012-10-19</td>\n",
       "      <td>1</td>\n",
       "      <td>24185.27</td>\n",
       "      <td>6293.257222</td>\n",
       "      <td>17892.012778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2012-10-26</td>\n",
       "      <td>1</td>\n",
       "      <td>27390.81</td>\n",
       "      <td>5727.487281</td>\n",
       "      <td>21663.322719</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        Date  Weight  Observed_Values  Predicted_Values       Derived\n",
       "0 2012-09-28       1         18947.81       6539.697097  12408.112903\n",
       "1 2012-10-05       1         21904.47       6539.697097  15364.772903\n",
       "2 2012-10-12       1         22764.01       6539.697097  16224.312903\n",
       "3 2012-10-19       1         24185.27       6293.257222  17892.012778\n",
       "4 2012-10-26       1         27390.81       5727.487281  21663.322719"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weight = np.where(test_set['IsHoliday_x']==0, 1, 5)\n",
    "\n",
    "df = pd.DataFrame({\"Date\": test_set['Date'],\n",
    "                   \"Weight\": weight,\n",
    "                   \"Observed_Values\":observed_weekly_sales.reshape(observed_weekly_sales.shape[0]),\n",
    "                  \"Predicted_Values\":predicted_weekly_sales.reshape(predicted_weekly_sales.shape[0])})\n",
    "\n",
    "df['Derived'] = df['Weight']*abs(df['Observed_Values']-df['Predicted_Values'])\n",
    "\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x2c86eaa7f98>]"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(df.Derived[0:10])\n",
    "plt.plot(df.Observed_Values[0:10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "12730.936111088342"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "WMAE = sum(df['Derived']) / sum(df['Weight'])\n",
    "WMAE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "228098500.22196758 561995677.199496\n"
     ]
    }
   ],
   "source": [
    "# calculate the accuray of the model by sum of Square and mean absolute prediction error\n",
    "MAPE = np.mean(abs( sum(df['Observed_Values'])- sum(df['Predicted_Values'])/sum(df['Observed_Values'])))\n",
    "MSSE = np.mean(np.square(df['Predicted_Values'] -df['Observed_Values']))\n",
    "print(MAPE, MSSE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['val_loss', 'val_acc', 'loss', 'acc'])"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history_dict = history.history\n",
    "history_dict.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(1, len(loss)+1)\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and Validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()\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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}