cc_lstm.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np \n",
    "import pandas as pd \n",
    "import matplotlib.pyplot as plt\n",
    "from math import sqrt\n",
    "from numpy import concatenate\n",
    "from pandas import read_csv\n",
    "from pandas import DataFrame\n",
    "from pandas import concat\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import LSTM\n",
    "\n",
    "%matplotlib inline\n",
    "plt.style.use('seaborn')\n",
    "from fbprophet import Prophet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = pd.read_csv('./Chicago_Crimes_2012_to_2017_new.csv', error_bad_lines=False,index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Dataset Shape before drop_duplicate : ', (1048575, 13))\n"
     ]
    }
   ],
   "source": [
    "print('Dataset Shape before drop_duplicate : ', dataset.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Primary_Type</th>\n",
       "      <th>Description</th>\n",
       "      <th>Location_Description</th>\n",
       "      <th>Arrest</th>\n",
       "      <th>Domestic</th>\n",
       "      <th>Beat</th>\n",
       "      <th>District</th>\n",
       "      <th>Ward</th>\n",
       "      <th>Community_Area</th>\n",
       "      <th>X_Coordinate</th>\n",
       "      <th>Y_Coordinate</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>Longitude</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Date</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5/3/16 23:40</th>\n",
       "      <td>BATTERY</td>\n",
       "      <td>DOMESTIC BATTERY SIMPLE</td>\n",
       "      <td>APARTMENT</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>1022</td>\n",
       "      <td>10.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>29.0</td>\n",
       "      <td>1154907.0</td>\n",
       "      <td>1893681.0</td>\n",
       "      <td>41.864073</td>\n",
       "      <td>-87.706819</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5/3/16 21:40</th>\n",
       "      <td>BATTERY</td>\n",
       "      <td>DOMESTIC BATTERY SIMPLE</td>\n",
       "      <td>RESIDENCE</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>313</td>\n",
       "      <td>3.0</td>\n",
       "      <td>20.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>1183066.0</td>\n",
       "      <td>1864330.0</td>\n",
       "      <td>41.782922</td>\n",
       "      <td>-87.604363</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5/3/16 23:31</th>\n",
       "      <td>PUBLIC PEACE VIOLATION</td>\n",
       "      <td>RECKLESS CONDUCT</td>\n",
       "      <td>STREET</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1524</td>\n",
       "      <td>15.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>1140789.0</td>\n",
       "      <td>1904819.0</td>\n",
       "      <td>41.894908</td>\n",
       "      <td>-87.758372</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5/3/16 22:10</th>\n",
       "      <td>BATTERY</td>\n",
       "      <td>SIMPLE</td>\n",
       "      <td>SIDEWALK</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>1532</td>\n",
       "      <td>15.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>1143223.0</td>\n",
       "      <td>1901475.0</td>\n",
       "      <td>41.885687</td>\n",
       "      <td>-87.749516</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5/3/16 22:00</th>\n",
       "      <td>THEFT</td>\n",
       "      <td>$500 AND UNDER</td>\n",
       "      <td>RESIDENCE</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>1523</td>\n",
       "      <td>15.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>1139890.0</td>\n",
       "      <td>1901675.0</td>\n",
       "      <td>41.886297</td>\n",
       "      <td>-87.761751</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                        Primary_Type              Description  \\\n",
       "Date                                                            \n",
       "5/3/16 23:40                 BATTERY  DOMESTIC BATTERY SIMPLE   \n",
       "5/3/16 21:40                 BATTERY  DOMESTIC BATTERY SIMPLE   \n",
       "5/3/16 23:31  PUBLIC PEACE VIOLATION         RECKLESS CONDUCT   \n",
       "5/3/16 22:10                 BATTERY                   SIMPLE   \n",
       "5/3/16 22:00                   THEFT           $500 AND UNDER   \n",
       "\n",
       "             Location_Description  Arrest  Domestic  Beat  District  Ward  \\\n",
       "Date                                                                        \n",
       "5/3/16 23:40            APARTMENT    True      True  1022      10.0  24.0   \n",
       "5/3/16 21:40            RESIDENCE   False      True   313       3.0  20.0   \n",
       "5/3/16 23:31               STREET   False     False  1524      15.0  37.0   \n",
       "5/3/16 22:10             SIDEWALK   False     False  1532      15.0  28.0   \n",
       "5/3/16 22:00            RESIDENCE   False      True  1523      15.0  28.0   \n",
       "\n",
       "              Community_Area  X_Coordinate  Y_Coordinate   Latitude  Longitude  \n",
       "Date                                                                            \n",
       "5/3/16 23:40            29.0     1154907.0     1893681.0  41.864073 -87.706819  \n",
       "5/3/16 21:40            42.0     1183066.0     1864330.0  41.782922 -87.604363  \n",
       "5/3/16 23:31            25.0     1140789.0     1904819.0  41.894908 -87.758372  \n",
       "5/3/16 22:10            25.0     1143223.0     1901475.0  41.885687 -87.749516  \n",
       "5/3/16 22:00            25.0     1139890.0     1901675.0  41.886297 -87.761751  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# dataset = dataset.head(100)\n",
    "dataset.head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# transform string data to number"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Primary_Type string to number\n",
    "unique_dict = dict((c, i) for i, c in enumerate(dataset['Primary_Type'].unique()))\n",
    "dataset['Primary_Type'] = [unique_dict[t] for d, t in dataset.Primary_Type.iteritems()]\n",
    "\n",
    "# Description string to number\n",
    "unique_dictx = dict((c, i) for i, c in enumerate(dataset['Description'].unique()))\n",
    "dataset['Description'] = [unique_dictx[t] for d, t in dataset.Description.iteritems()]\n",
    "\n",
    "# Location_Description string to number\n",
    "unique_dictx = dict((c, i) for i, c in enumerate(dataset['Location_Description'].unique()))\n",
    "dataset['Location_Description'] = [unique_dictx[t] for d, t in dataset.Location_Description.iteritems()]\n",
    "\n",
    "# Arrest, Domestic bool True/False to 0/1\n",
    "dataset.Arrest = dataset.Arrest.astype(int)\n",
    "dataset.Domestic = dataset.Domestic.astype(int)\n",
    "\n",
    "dataset.head()\n",
    "dataset.to_csv('cc12-17_numeric.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Index: 100 entries, 5/3/16 23:40 to 5/3/16 14:25\n",
      "Data columns (total 13 columns):\n",
      "Primary_Type            100 non-null int64\n",
      "Description             100 non-null int64\n",
      "Location_Description    100 non-null int64\n",
      "Arrest                  100 non-null int64\n",
      "Domestic                100 non-null int64\n",
      "Beat                    100 non-null int64\n",
      "District                100 non-null float64\n",
      "Ward                    100 non-null float64\n",
      "Community_Area          100 non-null float64\n",
      "X_Coordinate            100 non-null float64\n",
      "Y_Coordinate            100 non-null float64\n",
      "Latitude                100 non-null float64\n",
      "Longitude               100 non-null float64\n",
      "dtypes: float64(7), int64(6)\n",
      "memory usage: 10.9+ KB\n"
     ]
    }
   ],
   "source": [
    "dataset.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# convert series to supervised learning\n",
    "\n",
    "def series_to_supervised(data, n_in=1, n_out=1, dropnan=True):\n",
    "    n_vars = 1 if type(data) is list else data.shape[1]\n",
    "    df = DataFrame(data)\n",
    "    cols, names = list(), list()\n",
    "    # input sequence (t-n, ... t-1)\n",
    "    for i in range(n_in, 0, -1):\n",
    "        cols.append(df.shift(i))\n",
    "        names += [('var%d(t-%d)' % (j + 1, i)) for j in range(n_vars)]\n",
    "    # forecast sequence (t, t+1, ... t+n)\n",
    "    for i in range(0, n_out):\n",
    "        cols.append(df.shift(-i))  # print(df.shift(-i).head())\n",
    "        if i == 0:\n",
    "            names += [('var%d(t)' % (j + 1)) for j in range(n_vars)]\n",
    "        else:\n",
    "            names += [('var%d(t+%d)' % (j + 1, i)) for j in range(n_vars)]\n",
    "    # put it all together\n",
    "    agg = concat(cols, axis=1)\n",
    "    agg.columns = names\n",
    "    # drop rows with NaN values\n",
    "    if dropnan:\n",
    "        agg.dropna(inplace=True)\n",
    "    return agg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(100, 13)\n",
      "   var1(t-1)  var2(t-1)  var3(t-1)  var4(t-1)  var5(t-1)  var6(t-1)  \\\n",
      "1   0.000000   0.000000   0.000000        1.0        1.0   0.375825   \n",
      "2   0.000000   0.000000   0.045455        0.0        1.0   0.083333   \n",
      "3   0.083333   0.033333   0.090909        0.0        0.0   0.582921   \n",
      "4   0.000000   0.066667   0.136364        0.0        0.0   0.586221   \n",
      "5   0.166667   0.100000   0.045455        0.0        1.0   0.582508   \n",
      "\n",
      "   var7(t-1)  var8(t-1)  var9(t-1)  var10(t-1)    ...     var4(t)  var5(t)  \\\n",
      "1   0.375000   0.469388   0.368421    0.360952    ...         0.0      1.0   \n",
      "2   0.083333   0.387755   0.539474    0.767370    ...         0.0      0.0   \n",
      "3   0.583333   0.734694   0.315789    0.157188    ...         0.0      0.0   \n",
      "4   0.583333   0.551020   0.315789    0.192320    ...         0.0      1.0   \n",
      "5   0.583333   0.551020   0.315789    0.144215    ...         0.0      0.0   \n",
      "\n",
      "    var6(t)   var7(t)   var8(t)   var9(t)  var10(t)  var11(t)  var12(t)  \\\n",
      "1  0.083333  0.083333  0.387755  0.539474  0.767370  0.374900  0.374245   \n",
      "2  0.582921  0.583333  0.734694  0.315789  0.157188  0.696142  0.697464   \n",
      "3  0.586221  0.583333  0.551020  0.315789  0.192320  0.669611  0.670845   \n",
      "4  0.582508  0.583333  0.551020  0.315789  0.144215  0.671197  0.672615   \n",
      "5  0.214521  0.208333  0.142857  0.565789  0.771267  0.266299  0.265823   \n",
      "\n",
      "   var13(t)  \n",
      "1  0.767456  \n",
      "2  0.155579  \n",
      "3  0.190735  \n",
      "4  0.142151  \n",
      "5  0.769684  \n",
      "\n",
      "[5 rows x 26 columns]\n",
      "   var1(t-1)  var2(t-1)  var3(t-1)  var4(t-1)  var5(t-1)  var6(t-1)  \\\n",
      "1   0.000000   0.000000   0.000000        1.0        1.0   0.375825   \n",
      "2   0.000000   0.000000   0.045455        0.0        1.0   0.083333   \n",
      "3   0.083333   0.033333   0.090909        0.0        0.0   0.582921   \n",
      "4   0.000000   0.066667   0.136364        0.0        0.0   0.586221   \n",
      "5   0.166667   0.100000   0.045455        0.0        1.0   0.582508   \n",
      "\n",
      "   var7(t-1)  var8(t-1)  var9(t-1)  var10(t-1)  var11(t-1)  var12(t-1)  \\\n",
      "1   0.375000   0.469388   0.368421    0.360952    0.607773    0.608475   \n",
      "2   0.083333   0.387755   0.539474    0.767370    0.374900    0.374245   \n",
      "3   0.583333   0.734694   0.315789    0.157188    0.696142    0.697464   \n",
      "4   0.583333   0.551020   0.315789    0.192320    0.669611    0.670845   \n",
      "5   0.583333   0.551020   0.315789    0.144215    0.671197    0.672615   \n",
      "\n",
      "   var13(t-1)  var13(t)  \n",
      "1    0.360382  0.767456  \n",
      "2    0.767456  0.155579  \n",
      "3    0.155579  0.190735  \n",
      "4    0.190735  0.142151  \n",
      "5    0.142151  0.769684  \n"
     ]
    }
   ],
   "source": [
    "values = dataset.values \n",
    "\n",
    "print values.shape\n",
    "\n",
    "# integer encode direction\n",
    "encoder = LabelEncoder()\n",
    "\n",
    "# ensure all data is float\n",
    "values = values.astype('float32')\n",
    "\n",
    "# normalize features\n",
    "scaler = MinMaxScaler(feature_range=(0, 1))\n",
    "scaled = scaler.fit_transform(values)\n",
    "\n",
    "# print(scaled[:5])\n",
    "\n",
    "# frame as supervised learning\n",
    "reframed = series_to_supervised(scaled, 1, 1)\n",
    "\n",
    "# drop columns we don't want to predict\n",
    "print(reframed.head(5))\n",
    "reframed.drop(reframed.columns[[13,14,15,16,17,18,19,20,21,22,23,24]], axis=1, inplace=True)\n",
    "print(reframed.head(5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "((59, 1, 13), (59, 1), (40, 1, 13), (40, 1))\n"
     ]
    }
   ],
   "source": [
    "train=0.6\n",
    "test=0.4\n",
    "# split into train and test sets\n",
    "values = reframed.values\n",
    "n_train_hours = int(reframed.shape[0]*train)\n",
    "train = values[:n_train_hours, :]\n",
    "test = values[n_train_hours:, :]\n",
    "# split into input and outputs\n",
    "train_X, train_y = train[:, :-1], train[:, -1:]\n",
    "test_X, test_y = test[:, :-1], test[:, -1:]\n",
    "# reshape input to be 3D [samples, timesteps, features]\n",
    "train_X = train_X.reshape((train_X.shape[0], 1, train_X.shape[1]))\n",
    "test_X = test_X.reshape((test_X.shape[0], 1, test_X.shape[1]))\n",
    "print(train_X.shape, train_y.shape, test_X.shape, test_y.shape)\n",
    "#print(train_y[:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "lstm_1 (LSTM)                (None, 75)                26700     \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 100)               7600      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 101       \n",
      "=================================================================\n",
      "Total params: 34,401\n",
      "Trainable params: 34,401\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# design network\n",
    "model = Sequential()\n",
    "model.add(LSTM(75, input_shape=(train_X.shape[1], train_X.shape[2])))\n",
    "model.add(Dense(100))\n",
    "model.add(Dense(1))\n",
    "# using mean_absolute_error loss function\n",
    "model.compile(loss='mae', optimizer='rmsprop')\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 59 samples, validate on 40 samples\n",
      "Epoch 1/70\n",
      "59/59 [==============================] - 1s - loss: 0.4530 - val_loss: 0.2404\n",
      "Epoch 2/70\n",
      "59/59 [==============================] - 0s - loss: 0.2614 - val_loss: 0.2023\n",
      "Epoch 3/70\n",
      "59/59 [==============================] - 0s - loss: 0.1839 - val_loss: 0.2110\n",
      "Epoch 4/70\n",
      "59/59 [==============================] - 0s - loss: 0.1782 - val_loss: 0.2081\n",
      "Epoch 5/70\n",
      "59/59 [==============================] - 0s - loss: 0.1782 - val_loss: 0.2043\n",
      "Epoch 6/70\n",
      "59/59 [==============================] - 0s - loss: 0.1776 - val_loss: 0.2125\n",
      "Epoch 7/70\n",
      "59/59 [==============================] - 0s - loss: 0.1750 - val_loss: 0.2031\n",
      "Epoch 8/70\n",
      "59/59 [==============================] - 0s - loss: 0.1756 - val_loss: 0.2047\n",
      "Epoch 9/70\n",
      "59/59 [==============================] - 0s - loss: 0.1746 - val_loss: 0.1984\n",
      "Epoch 10/70\n",
      "59/59 [==============================] - 0s - loss: 0.1752 - val_loss: 0.2029\n",
      "Epoch 11/70\n",
      "59/59 [==============================] - 0s - loss: 0.1726 - val_loss: 0.1962\n",
      "Epoch 12/70\n",
      "59/59 [==============================] - 0s - loss: 0.1734 - val_loss: 0.2124\n",
      "Epoch 13/70\n",
      "59/59 [==============================] - 0s - loss: 0.1690 - val_loss: 0.1962\n",
      "Epoch 14/70\n",
      "59/59 [==============================] - 0s - loss: 0.1711 - val_loss: 0.1938\n",
      "Epoch 15/70\n",
      "59/59 [==============================] - 0s - loss: 0.1709 - val_loss: 0.2144\n",
      "Epoch 16/70\n",
      "59/59 [==============================] - 0s - loss: 0.1668 - val_loss: 0.2318\n",
      "Epoch 17/70\n",
      "59/59 [==============================] - 0s - loss: 0.1687 - val_loss: 0.2097\n",
      "Epoch 18/70\n",
      "59/59 [==============================] - 0s - loss: 0.1652 - val_loss: 0.2148\n",
      "Epoch 19/70\n",
      "59/59 [==============================] - 0s - loss: 0.1656 - val_loss: 0.1922\n",
      "Epoch 20/70\n",
      "59/59 [==============================] - 0s - loss: 0.1710 - val_loss: 0.1929\n",
      "Epoch 21/70\n",
      "59/59 [==============================] - 0s - loss: 0.1754 - val_loss: 0.2141\n",
      "Epoch 22/70\n",
      "59/59 [==============================] - 0s - loss: 0.1648 - val_loss: 0.1883\n",
      "Epoch 23/70\n",
      "59/59 [==============================] - 0s - loss: 0.1687 - val_loss: 0.2141\n",
      "Epoch 24/70\n",
      "59/59 [==============================] - 0s - loss: 0.1648 - val_loss: 0.2338\n",
      "Epoch 25/70\n",
      "59/59 [==============================] - 0s - loss: 0.1712 - val_loss: 0.2442\n",
      "Epoch 26/70\n",
      "59/59 [==============================] - 0s - loss: 0.1714 - val_loss: 0.2338\n",
      "Epoch 27/70\n",
      "59/59 [==============================] - 0s - loss: 0.1664 - val_loss: 0.1941\n",
      "Epoch 28/70\n",
      "59/59 [==============================] - 0s - loss: 0.1661 - val_loss: 0.1869\n",
      "Epoch 29/70\n",
      "59/59 [==============================] - 0s - loss: 0.1680 - val_loss: 0.2186\n",
      "Epoch 30/70\n",
      "59/59 [==============================] - 0s - loss: 0.1642 - val_loss: 0.1925\n",
      "Epoch 31/70\n",
      "59/59 [==============================] - 0s - loss: 0.1657 - val_loss: 0.1880\n",
      "Epoch 32/70\n",
      "59/59 [==============================] - 0s - loss: 0.1706 - val_loss: 0.2171\n",
      "Epoch 33/70\n",
      "59/59 [==============================] - 0s - loss: 0.1642 - val_loss: 0.2269\n",
      "Epoch 34/70\n",
      "59/59 [==============================] - 0s - loss: 0.1693 - val_loss: 0.2375\n",
      "Epoch 35/70\n",
      "59/59 [==============================] - 0s - loss: 0.1700 - val_loss: 0.2379\n",
      "Epoch 36/70\n",
      "59/59 [==============================] - 0s - loss: 0.1682 - val_loss: 0.2133\n",
      "Epoch 37/70\n",
      "59/59 [==============================] - 0s - loss: 0.1643 - val_loss: 0.1893\n",
      "Epoch 38/70\n",
      "59/59 [==============================] - 0s - loss: 0.1654 - val_loss: 0.2060\n",
      "Epoch 39/70\n",
      "59/59 [==============================] - 0s - loss: 0.1624 - val_loss: 0.1902\n",
      "Epoch 40/70\n",
      "59/59 [==============================] - 0s - loss: 0.1646 - val_loss: 0.1886\n",
      "Epoch 41/70\n",
      "59/59 [==============================] - 0s - loss: 0.1706 - val_loss: 0.2231\n",
      "Epoch 42/70\n",
      "59/59 [==============================] - 0s - loss: 0.1677 - val_loss: 0.2333\n",
      "Epoch 43/70\n",
      "59/59 [==============================] - 0s - loss: 0.1688 - val_loss: 0.2337\n",
      "Epoch 44/70\n",
      "59/59 [==============================] - 0s - loss: 0.1671 - val_loss: 0.2108\n",
      "Epoch 45/70\n",
      "59/59 [==============================] - 0s - loss: 0.1632 - val_loss: 0.1880\n",
      "Epoch 46/70\n",
      "59/59 [==============================] - 0s - loss: 0.1647 - val_loss: 0.2025\n",
      "Epoch 47/70\n",
      "59/59 [==============================] - 0s - loss: 0.1634 - val_loss: 0.1882\n",
      "Epoch 48/70\n",
      "59/59 [==============================] - 0s - loss: 0.1689 - val_loss: 0.2107\n",
      "Epoch 49/70\n",
      "59/59 [==============================] - 0s - loss: 0.1624 - val_loss: 0.1977\n",
      "Epoch 50/70\n",
      "59/59 [==============================] - 0s - loss: 0.1659 - val_loss: 0.1889\n",
      "Epoch 51/70\n",
      "59/59 [==============================] - ETA: 0s - loss: 0.193 - 0s - loss: 0.1695 - val_loss: 0.2142\n",
      "Epoch 52/70\n",
      "59/59 [==============================] - 0s - loss: 0.1649 - val_loss: 0.2239\n",
      "Epoch 53/70\n",
      "59/59 [==============================] - 0s - loss: 0.1657 - val_loss: 0.2354\n",
      "Epoch 54/70\n",
      "59/59 [==============================] - 0s - loss: 0.1684 - val_loss: 0.2105\n",
      "Epoch 55/70\n",
      "59/59 [==============================] - 0s - loss: 0.1618 - val_loss: 0.1901\n",
      "Epoch 56/70\n",
      "59/59 [==============================] - 0s - loss: 0.1630 - val_loss: 0.1896\n",
      "Epoch 57/70\n",
      "59/59 [==============================] - 0s - loss: 0.1626 - val_loss: 0.1880\n",
      "Epoch 58/70\n",
      "59/59 [==============================] - 0s - loss: 0.1662 - val_loss: 0.2049\n",
      "Epoch 59/70\n",
      "59/59 [==============================] - 0s - loss: 0.1642 - val_loss: 0.2367\n",
      "Epoch 60/70\n",
      "59/59 [==============================] - 0s - loss: 0.1725 - val_loss: 0.2259\n",
      "Epoch 61/70\n",
      "59/59 [==============================] - 0s - loss: 0.1665 - val_loss: 0.2267\n",
      "Epoch 62/70\n",
      "59/59 [==============================] - 0s - loss: 0.1657 - val_loss: 0.2054\n",
      "Epoch 63/70\n",
      "59/59 [==============================] - 0s - loss: 0.1610 - val_loss: 0.1909\n",
      "Epoch 64/70\n",
      "59/59 [==============================] - 0s - loss: 0.1622 - val_loss: 0.1872\n",
      "Epoch 65/70\n",
      "59/59 [==============================] - 0s - loss: 0.1630 - val_loss: 0.2069\n",
      "Epoch 66/70\n",
      "59/59 [==============================] - 0s - loss: 0.1623 - val_loss: 0.2390\n",
      "Epoch 67/70\n",
      "59/59 [==============================] - 0s - loss: 0.1727 - val_loss: 0.2191\n",
      "Epoch 68/70\n",
      "59/59 [==============================] - 0s - loss: 0.1635 - val_loss: 0.1960\n",
      "Epoch 69/70\n",
      "59/59 [==============================] - 0s - loss: 0.1632 - val_loss: 0.1901\n",
      "Epoch 70/70\n",
      "59/59 [==============================] - 0s - loss: 0.1661 - val_loss: 0.2111\n"
     ]
    }
   ],
   "source": [
    "# fit network\n",
    "history = model.fit(train_X, train_y, epochs=70, batch_size=32, validation_data=(test_X, test_y), verbose=1, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(40, 1, 13)\n"
     ]
    }
   ],
   "source": [
    "test_X = test_X.reshape((test_X.shape[0], 1, test_X.shape[2]))\n",
    "print(test_X.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.66418147]\n",
      " [ 0.65400076]\n",
      " [ 0.58043003]\n",
      " [ 0.63376999]\n",
      " [ 0.60490775]]\n",
      "[[ 0.37057495]\n",
      " [ 0.30105591]\n",
      " [ 0.67575073]\n",
      " [ 0.68212891]\n",
      " [ 0.60794067]\n",
      " [ 0.01641846]\n",
      " [ 0.70611572]\n",
      " [ 0.83123779]\n",
      " [ 0.3682251 ]\n",
      " [ 0.13537598]\n",
      " [ 0.68225098]\n",
      " [ 0.52337646]\n",
      " [ 0.67056274]\n",
      " [ 0.75372314]\n",
      " [ 0.44430542]\n",
      " [ 0.50219727]\n",
      " [ 0.58724976]\n",
      " [ 0.29693604]\n",
      " [ 0.19805908]\n",
      " [ 0.81399536]\n",
      " [ 0.00421143]\n",
      " [ 0.3644104 ]\n",
      " [ 0.87896729]\n",
      " [ 0.50265503]\n",
      " [ 0.67355347]\n",
      " [ 0.3291626 ]\n",
      " [ 0.37298584]\n",
      " [ 0.77258301]\n",
      " [ 0.34701538]\n",
      " [ 0.47686768]\n",
      " [ 0.30343628]\n",
      " [ 0.76174927]\n",
      " [ 0.12121582]\n",
      " [ 0.87127686]\n",
      " [ 0.26898193]\n",
      " [ 0.45199585]\n",
      " [ 0.16680908]\n",
      " [ 0.08215332]\n",
      " [ 0.57049561]\n",
      " [ 0.53128052]]\n",
      "[ 7.97017765  7.84800911  6.96516037  7.60523987  7.25889301]\n",
      "[ 4.44689941  3.6126709   8.10900879  8.18554688  7.29528809]\n",
      "Test RMSE: 3.185\n"
     ]
    }
   ],
   "source": [
    "# make a prediction\n",
    "yhat = model.predict(test_X)\n",
    "print(yhat[:5])\n",
    "test_X = test_X.reshape((test_X.shape[0], test_X.shape[2]))\n",
    "print(test_y[:,-1:])\n",
    "# invert scaling for forecast\n",
    "inv_yhat = concatenate((yhat, test_X[:, 1:]), axis=1)\n",
    "inv_yhat = scaler.inverse_transform(inv_yhat)\n",
    "inv_yhat = inv_yhat[:,0]\n",
    "print(inv_yhat[:5])\n",
    "# invert scaling for actual\n",
    "test_y = test_y.reshape((len(test_y), 1))\n",
    "inv_y = concatenate((test_y, test_X[:, 1:]), axis=1)\n",
    "inv_y = scaler.inverse_transform(inv_y)\n",
    "inv_y = inv_y[:,0]\n",
    "print(inv_y[:5])\n",
    "# calculate RMSE\n",
    "rmse = sqrt(mean_squared_error(inv_y, inv_yhat))\n",
    "print('Test RMSE: %.3f' % rmse)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}