Chapter 25 Time series prediction with LSTM.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas\n",
    "import math\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import LSTM\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.metrics import mean_squared_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_dataset(dataset,look_back=1):\n",
    "    dataX, dataY = [],[]\n",
    "    for i in range(len(dataset)-look_back-1):\n",
    "        a = dataset[i:(i+look_back),0]\n",
    "        dataX.append(a)\n",
    "        dataY.append(dataset[i+look_back,0])\n",
    "    return numpy.array(dataX),numpy.array(dataY)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "numpy.random.seed(7)\n",
    "dataframe = pandas.read_csv(\"AirPassengers.csv\",usecols=[1],engine=\"python\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = dataframe.values\n",
    "dataset = dataset.astype('float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = MinMaxScaler(feature_range=(0,1))\n",
    "dataset = scaler.fit_transform(dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_size = int(len(dataset)*0.67)\n",
    "test_size = len(dataset)-train_size\n",
    "train,test = dataset[0:train_size,:],dataset[train_size:len(dataset),:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "look_back = 1\n",
    "trainX,trainY = create_dataset(train,look_back)\n",
    "testX,testY = create_dataset(test,look_back)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainX = numpy.reshape(trainX,(trainX.shape[0],1,trainX.shape[1]))\n",
    "testX = numpy.reshape(testX,(testX.shape[0],1,testX.shape[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/arvind/.local/lib/python3.6/site-packages/ipykernel_launcher.py:2: UserWarning: The `input_dim` and `input_length` arguments in recurrent layers are deprecated. Use `input_shape` instead.\n",
      "  \n",
      "/home/arvind/.local/lib/python3.6/site-packages/ipykernel_launcher.py:2: UserWarning: Update your `LSTM` call to the Keras 2 API: `LSTM(4, input_shape=(None, 1))`\n",
      "  \n",
      "W0426 19:32:05.645480 140039748347712 deprecation.py:506] From /home/arvind/.local/lib/python3.6/site-packages/tensorflow_core/python/ops/resource_variable_ops.py:1633: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "If using Keras pass *_constraint arguments to layers.\n",
      "W0426 19:32:06.648608 140039748347712 deprecation.py:323] From /home/arvind/.local/lib/python3.6/site-packages/keras/optimizers.py:550: BaseResourceVariable.constraint (from tensorflow.python.ops.resource_variable_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Apply a constraint manually following the optimizer update step.\n",
      "W0426 19:32:06.864027 140039748347712 module_wrapper.py:136] From /home/arvind/.local/lib/python3.6/site-packages/tensorflow_core/python/util/module_wrapper.py:163: The name tf.global_variables is deprecated. Please use tf.compat.v1.global_variables instead.\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      " - 1s - loss: 0.0410\n",
      "Epoch 2/100\n",
      " - 0s - loss: 0.0197\n",
      "Epoch 3/100\n",
      " - 0s - loss: 0.0142\n",
      "Epoch 4/100\n",
      " - 0s - loss: 0.0127\n",
      "Epoch 5/100\n",
      " - 0s - loss: 0.0117\n",
      "Epoch 6/100\n",
      " - 0s - loss: 0.0106\n",
      "Epoch 7/100\n",
      " - 0s - loss: 0.0096\n",
      "Epoch 8/100\n",
      " - 0s - loss: 0.0087\n",
      "Epoch 9/100\n",
      " - 0s - loss: 0.0076\n",
      "Epoch 10/100\n",
      " - 0s - loss: 0.0065\n",
      "Epoch 11/100\n",
      " - 0s - loss: 0.0057\n",
      "Epoch 12/100\n",
      " - 0s - loss: 0.0048\n",
      "Epoch 13/100\n",
      " - 0s - loss: 0.0041\n",
      "Epoch 14/100\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 15/100\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 16/100\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 17/100\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 18/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 19/100\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 20/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 21/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 22/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 23/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 24/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 25/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 26/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 27/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 28/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 29/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 30/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 31/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 32/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 33/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 34/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 35/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 36/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 37/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 38/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 39/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 40/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 41/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 42/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 43/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 44/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 45/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 46/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 47/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 48/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 49/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 50/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 51/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 52/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 53/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 54/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 55/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 56/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 57/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 58/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 59/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 60/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 61/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 62/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 63/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 64/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 65/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 66/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 67/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 68/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 69/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 70/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 71/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 72/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 73/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 74/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 75/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 76/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 77/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 78/100\n",
      " - 0s - loss: 0.0019\n",
      "Epoch 79/100\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 80/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 81/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 82/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 83/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 84/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 85/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 86/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 87/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 88/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 89/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 90/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 91/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 92/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 93/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 94/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 95/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 96/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 97/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 98/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 99/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 100/100\n",
      " - 0s - loss: 0.0020\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.callbacks.History at 0x7f5d2c482518>"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(LSTM(4,input_dim=look_back))\n",
    "model.add(Dense(1))\n",
    "model.compile(loss='mean_squared_error',optimizer='adam')\n",
    "model.fit(trainX,trainY,epochs=100,batch_size=1,verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredict = model.predict(trainX)\n",
    "testPredict = model.predict(testX)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredict = scaler.inverse_transform(trainPredict)\n",
    "trainY = scaler.inverse_transform([trainY])\n",
    "testPredict = scaler.inverse_transform(testPredict)\n",
    "testY = scaler.inverse_transform([testY])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Score: 22.93 RMSE\n"
     ]
    }
   ],
   "source": [
    "trainScore = math.sqrt(mean_squared_error(trainY[0],trainPredict[:,0]))\n",
    "print(\"Train Score: %.2f RMSE\"%(trainScore))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Score: 47.60 RMSE\n"
     ]
    }
   ],
   "source": [
    "testScore = math.sqrt(mean_squared_error(testY[0],testPredict[:,0]))\n",
    "print(\"Test Score: %.2f RMSE\"%(testScore))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredictPlot = numpy.empty_like(dataset)\n",
    "trainPredictPlot[:,:] = numpy.nan\n",
    "trainPredictPlot[look_back:len(trainPredict)+look_back,:] = trainPredict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "testPredictPlot = numpy.empty_like(dataset)\n",
    "testPredictPlot[:,:] = numpy.nan\n",
    "testPredictPlot[len(trainPredict)+(look_back*2)+1:len(dataset)-1,:] = testPredict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "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(scaler.inverse_transform(dataset))\n",
    "plt.plot(trainPredictPlot)\n",
    "plt.plot(testPredictPlot)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "#LSTM for regression using window method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "look_back = 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainX, trainY = create_dataset(train,look_back)\n",
    "testX, testY = create_dataset(test,look_back)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainX = numpy.reshape(trainX,(trainX.shape[0],1,trainX.shape[1]))\n",
    "testX = numpy.reshape(testX,(testX.shape[0],1,testX.shape[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/arvind/.local/lib/python3.6/site-packages/ipykernel_launcher.py:2: UserWarning: The `input_dim` and `input_length` arguments in recurrent layers are deprecated. Use `input_shape` instead.\n",
      "  \n",
      "/home/arvind/.local/lib/python3.6/site-packages/ipykernel_launcher.py:2: UserWarning: Update your `LSTM` call to the Keras 2 API: `LSTM(4, input_shape=(None, 3))`\n",
      "  \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      " - 0s - loss: 0.0515\n",
      "Epoch 2/100\n",
      " - 0s - loss: 0.0242\n",
      "Epoch 3/100\n",
      " - 0s - loss: 0.0157\n",
      "Epoch 4/100\n",
      " - 0s - loss: 0.0134\n",
      "Epoch 5/100\n",
      " - 0s - loss: 0.0120\n",
      "Epoch 6/100\n",
      " - 0s - loss: 0.0106\n",
      "Epoch 7/100\n",
      " - 0s - loss: 0.0094\n",
      "Epoch 8/100\n",
      " - 0s - loss: 0.0082\n",
      "Epoch 9/100\n",
      " - 0s - loss: 0.0072\n",
      "Epoch 10/100\n",
      " - 0s - loss: 0.0063\n",
      "Epoch 11/100\n",
      " - 0s - loss: 0.0056\n",
      "Epoch 12/100\n",
      " - 0s - loss: 0.0050\n",
      "Epoch 13/100\n",
      " - 0s - loss: 0.0046\n",
      "Epoch 14/100\n",
      " - 0s - loss: 0.0043\n",
      "Epoch 15/100\n",
      " - 0s - loss: 0.0041\n",
      "Epoch 16/100\n",
      " - 0s - loss: 0.0039\n",
      "Epoch 17/100\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 18/100\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 19/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 20/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 21/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 22/100\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 23/100\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 24/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 25/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 26/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 27/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 28/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 29/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 30/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 31/100\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 32/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 33/100\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 34/100\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 35/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 36/100\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 37/100\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 38/100\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 39/100\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 40/100\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 41/100\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 42/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 43/100\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 44/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 45/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 46/100\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 47/100\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 48/100\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 49/100\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 50/100\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 51/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 52/100\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 53/100\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 54/100\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 55/100\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 56/100\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 57/100\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 58/100\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 59/100\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 60/100\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 61/100\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 62/100\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 63/100\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 64/100\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 65/100\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 66/100\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 67/100\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 68/100\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 69/100\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 70/100\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 71/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 72/100\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 73/100\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 74/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 75/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 76/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 77/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 78/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 79/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 80/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 81/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 82/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 83/100\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 84/100\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 85/100\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 86/100\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 87/100\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 88/100\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 89/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 90/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 91/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 92/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 93/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 94/100\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 95/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 96/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 97/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 98/100\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 99/100\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 100/100\n",
      " - 0s - loss: 0.0021\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.callbacks.History at 0x7f5d240dd828>"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(LSTM(4,input_dim=look_back))\n",
    "model.add(Dense(1))\n",
    "model.compile(loss = \"mean_squared_error\",optimizer=\"adam\")\n",
    "model.fit(trainX,trainY,epochs=100,batch_size=1,verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredict = model.predict(trainX)\n",
    "testPredict = model.predict(testX)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredict = scaler.inverse_transform(trainPredict)\n",
    "trainY = scaler.inverse_transform([trainY])\n",
    "testPredict = scaler.inverse_transform(testPredict)\n",
    "testY = scaler.inverse_transform([testY])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainScore = math.sqrt(mean_squared_error(trainY[0],trainPredict[:,0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Score: 23.67 RMSE\n"
     ]
    }
   ],
   "source": [
    "print(\"Train Score: %.2f RMSE\"%(trainScore))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredictPlot = numpy.empty_like(dataset)\n",
    "trainPredictPlot[:,:] = numpy.nan\n",
    "trainPredictPlot[look_back:len(trainPredict)+look_back,:]=trainPredict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "testPredictPlot = numpy.empty_like(dataset)\n",
    "testPredictPlot[:,:] = numpy.nan\n",
    "testPredictPlot[len(trainPredict)+look_back*2+1:len(dataset)-1,:]=testPredict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "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(scaler.inverse_transform(dataset))\n",
    "plt.plot(trainPredictPlot)\n",
    "plt.plot(testPredictPlot)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "# LSTM for regression with time steps"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainX, trainY = create_dataset(train,look_back)\n",
    "testX, testY = create_dataset(test,look_back)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainX = numpy.reshape(trainX,(trainX.shape[0],trainX.shape[1],1))\n",
    "testX = numpy.reshape(testX,(testX.shape[0],testX.shape[1],1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/arvind/.local/lib/python3.6/site-packages/ipykernel_launcher.py:2: UserWarning: The `input_dim` and `input_length` arguments in recurrent layers are deprecated. Use `input_shape` instead.\n",
      "  \n",
      "/home/arvind/.local/lib/python3.6/site-packages/ipykernel_launcher.py:2: UserWarning: Update your `LSTM` call to the Keras 2 API: `LSTM(4, input_shape=(None, 1))`\n",
      "  \n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      " - 1s - loss: 0.0450\n",
      "Epoch 2/100\n",
      " - 0s - loss: 0.0176\n",
      "Epoch 3/100\n",
      " - 0s - loss: 0.0129\n",
      "Epoch 4/100\n",
      " - 0s - loss: 0.0112\n",
      "Epoch 5/100\n",
      " - 0s - loss: 0.0096\n",
      "Epoch 6/100\n",
      " - 0s - loss: 0.0081\n",
      "Epoch 7/100\n",
      " - 0s - loss: 0.0069\n",
      "Epoch 8/100\n",
      " - 0s - loss: 0.0060\n",
      "Epoch 9/100\n",
      " - 0s - loss: 0.0051\n",
      "Epoch 10/100\n",
      " - 0s - loss: 0.0047\n",
      "Epoch 11/100\n",
      " - 0s - loss: 0.0045\n",
      "Epoch 12/100\n",
      " - 0s - loss: 0.0041\n",
      "Epoch 13/100\n",
      " - 0s - loss: 0.0040\n",
      "Epoch 14/100\n",
      " - 0s - loss: 0.0040\n",
      "Epoch 15/100\n",
      " - 0s - loss: 0.0039\n",
      "Epoch 16/100\n",
      " - 0s - loss: 0.0039\n",
      "Epoch 17/100\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 18/100\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 19/100\n",
      " - 0s - loss: 0.0039\n",
      "Epoch 20/100\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 21/100\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 22/100\n",
      " - 0s - loss: 0.0039\n",
      "Epoch 23/100\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 24/100\n",
      " - 0s - loss: 0.0037\n",
      "Epoch 25/100\n",
      " - 0s - loss: 0.0037\n",
      "Epoch 26/100\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 27/100\n",
      " - 0s - loss: 0.0037\n",
      "Epoch 28/100\n",
      " - 0s - loss: 0.0037\n",
      "Epoch 29/100\n",
      " - 0s - loss: 0.0037\n",
      "Epoch 30/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 31/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 32/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 33/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 34/100\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 35/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 36/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 37/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 38/100\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 39/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 40/100\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 41/100\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 42/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 43/100\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 44/100\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 45/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 46/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 47/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 48/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 49/100\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 50/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 51/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 52/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 53/100\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 54/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 55/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 56/100\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 57/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 58/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 59/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 60/100\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 61/100\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 62/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 63/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 64/100\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 65/100\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 66/100\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 67/100\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 68/100\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 69/100\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 70/100\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 71/100\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 72/100\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 73/100\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 74/100\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 75/100\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 76/100\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 77/100\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 78/100\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 79/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 80/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 81/100\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 82/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 83/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 84/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 85/100\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 86/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 87/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 88/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 89/100\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 90/100\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 91/100\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 92/100\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 93/100\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 94/100\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 95/100\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 96/100\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 97/100\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 98/100\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 99/100\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 100/100\n",
      " - 0s - loss: 0.0026\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.callbacks.History at 0x7f5d14ee4e48>"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(LSTM(4,input_dim=1))\n",
    "model.add(Dense(1))\n",
    "model.compile(loss=\"mean_squared_error\",optimizer=\"adam\")\n",
    "model.fit(trainX,trainY,epochs=100,batch_size=1,verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredict = model.predict(trainX)\n",
    "testPredict = model.predict(testX)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredict = scaler.inverse_transform(trainPredict)\n",
    "trainY = scaler.inverse_transform([trainY])\n",
    "testPredict = scaler.inverse_transform(testPredict)\n",
    "testY = scaler.inverse_transform([testY])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Score: 28.03 RMSE\n"
     ]
    }
   ],
   "source": [
    "trainScore = math.sqrt(mean_squared_error(trainY[0],trainPredict[:,0]))\n",
    "print(\"Train Score: %.2f RMSE\" %(trainScore))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Score: 71.28 RMSE\n"
     ]
    }
   ],
   "source": [
    "testScore = math.sqrt(mean_squared_error(testY[0],testPredict[:,0]))\n",
    "print(\"Test Score: %.2f RMSE\" %(testScore))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredictPlot = numpy.empty_like(dataset)\n",
    "trainPredictPlot[:,:] = numpy.nan\n",
    "trainPredictPlot[look_back:len(trainPredict)+look_back,:] = trainPredict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "testPredictPlot = numpy.empty_like(dataset)\n",
    "testPredictPlot[:,:] = numpy.nan\n",
    "testPredictPlot[len(trainPredict)+(look_back*2)+1:len(dataset)-1,:] = testPredict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "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(scaler.inverse_transform(dataset))\n",
    "plt.plot(trainPredictPlot)\n",
    "plt.plot(testPredictPlot)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [],
   "source": [
    "# LSTM with memory between batches"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas\n",
    "import math\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import LSTM\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.metrics import mean_squared_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_dataset(dataset,look_back=1):\n",
    "    dataX, dataY = [], []\n",
    "    for i in range(len(dataset)-look_back-1):\n",
    "        a = dataset[i:(i+look_back),0]\n",
    "        dataX.append(a)\n",
    "        dataY.append(dataset[i+look_back,0])\n",
    "    return numpy.array(dataX), numpy.array(dataY)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [],
   "source": [
    "numpy.random.seed(7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataframe = pandas.read_csv('AirPassengers.csv',usecols=[1],engine=\"python\")\n",
    "dataset = dataframe.values\n",
    "dataset = dataset.astype('float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = MinMaxScaler(feature_range=(0,1))\n",
    "dataset = scaler.fit_transform(dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_size = int(len(dataset)*0.67)\n",
    "test_size = len(dataset) - train_size\n",
    "train, test = dataset[0:train_size,:], dataset[train_size:len(dataset),:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [],
   "source": [
    "look_back=3\n",
    "trainX,trainY = create_dataset(train, look_back)\n",
    "testX, testY = create_dataset(test, look_back)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainX = numpy.reshape(trainX,(trainX.shape[0], trainX.shape[1],1))\n",
    "testX = numpy.reshape(testX,(testX.shape[0],testX.shape[1],1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [],
   "source": [
    "batch_size=1\n",
    "model = Sequential()\n",
    "model.add(LSTM(4,batch_input_shape=(batch_size,look_back,1),stateful = True))\n",
    "model.add(Dense(1))\n",
    "model.compile(loss=\"mean_squared_error\",optimizer=\"adam\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/1\n",
      " - 0s - loss: 0.0046\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0078\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0052\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0047\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0048\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0048\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0047\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0046\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0045\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0044\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0044\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0043\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0042\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0042\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0041\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0041\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0041\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0040\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0040\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0039\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0039\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0037\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0019\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0019\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0019\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0019\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0019\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0018\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0018\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0018\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0018\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0018\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0018\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0018\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0018\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0017\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0016\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0016\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0016\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0016\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0016\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0016\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0016\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0016\n"
     ]
    }
   ],
   "source": [
    "for i in range(100):\n",
    "    model.fit(trainX,trainY,epochs=1,batch_size=batch_size,verbose=2,shuffle = False)\n",
    "    model.reset_states()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredict = model.predict(trainX,batch_size=batch_size)\n",
    "model.reset_states()\n",
    "testPredict = model.predict(testX,batch_size=batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredict = scaler.inverse_transform(trainPredict)\n",
    "trainY = scaler.inverse_transform([trainY])\n",
    "testPredict = scaler.inverse_transform(testPredict)\n",
    "testY = scaler.inverse_transform([testY])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Score: 20.57 rmse\n"
     ]
    }
   ],
   "source": [
    "trainScore=math.sqrt(mean_squared_error(trainY[0],trainPredict[:,0]))\n",
    "print(\"Train Score: %.2F rmse\" % (trainScore))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Score: 52.74 RMSE\n"
     ]
    }
   ],
   "source": [
    "testScore = math.sqrt(mean_squared_error(testY[0],testPredict[:,0]))\n",
    "print(\"Test Score: %.2f RMSE\"%(testScore))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredictPlot = numpy.empty_like(dataset)\n",
    "trainPredictPlot[:,:] = numpy.nan\n",
    "trainPredictPlot[look_back:len(trainPredict)+look_back,:] = trainPredict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "testPredictPlot = numpy.empty_like(dataset)\n",
    "testPredictPlot[:,:] = numpy.nan\n",
    "testPredictPlot[len(trainPredict)+(look_back*2)+1:len(dataset)-1,:] = testPredict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "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(scaler.inverse_transform(dataset))\n",
    "plt.plot(trainPredictPlot)\n",
    "plt.plot(testPredictPlot)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Stacked LSTM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [],
   "source": [
    "look_back=3\n",
    "trainX,trainY = create_dataset(train, look_back)\n",
    "testX, testY = create_dataset(test, look_back)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainX = numpy.reshape(trainX,(trainX.shape[0], trainX.shape[1],1))\n",
    "testX = numpy.reshape(testX,(testX.shape[0],testX.shape[1],1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/1\n",
      " - 1s - loss: 0.0048\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0084\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0063\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0057\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0054\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0053\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0052\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0051\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0051\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0051\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0050\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0050\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0050\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0050\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0049\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0049\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0049\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0048\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0048\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0047\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0046\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0046\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0045\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0044\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0043\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0042\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0041\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0040\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0039\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0038\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0037\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0036\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0035\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0034\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0033\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0032\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0031\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0030\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0029\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0028\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0027\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0026\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0025\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0024\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0023\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0022\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0021\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0020\n",
      "Epoch 1/1\n",
      " - 0s - loss: 0.0020\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(LSTM(4,batch_input_shape=(batch_size,look_back,1),stateful=True,return_sequences=True))\n",
    "model.add(LSTM(4,batch_input_shape=(batch_size,look_back,1),stateful=True))\n",
    "model.add(Dense(1))\n",
    "model.compile(loss=\"mean_squared_error\",optimizer=\"adam\")\n",
    "for i in range(100):\n",
    "    model.fit(trainX,trainY,epochs=1,batch_size=batch_size,verbose=2,shuffle = False)\n",
    "    model.reset_states()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredict=model.predict(trainX,batch_size=batch_size)\n",
    "model.reset_states()\n",
    "testPredict = model.predict(testX,batch_size=batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredict=scaler.inverse_transform(trainPredict)\n",
    "trainY = scaler.inverse_transform([trainY])\n",
    "testPredict = scaler.inverse_transform(testPredict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {},
   "outputs": [],
   "source": [
    "testY = scaler.inverse_transform([testY])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Score 22.54 RMSE\n"
     ]
    }
   ],
   "source": [
    "trainScore = math.sqrt(mean_squared_error(trainY[0],trainPredict[:,0]))\n",
    "print(\"Train Score %.2f RMSE\" %(trainScore))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Score 93.99 RMSE\n"
     ]
    }
   ],
   "source": [
    "testScore = math.sqrt(mean_squared_error(testY[0],testPredict[:,0]))\n",
    "print(\"Test Score %.2f RMSE\" %(testScore))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "metadata": {},
   "outputs": [],
   "source": [
    "trainPredictPlot = numpy.empty_like(dataset)\n",
    "trainPredictPlot[:,:] = numpy.nan\n",
    "trainPredictPlot[look_back:len(trainPredict)+look_back,:] = trainPredict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [],
   "source": [
    "testPredictPlot = numpy.empty_like(dataset)\n",
    "testPredictPlot[:,:] = numpy.nan\n",
    "testPredictPlot[len(trainPredict)+(look_back*2)+1:len(dataset)-1,:] = testPredict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "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(scaler.inverse_transform(dataset))\n",
    "plt.plot(trainPredictPlot)\n",
    "plt.plot(testPredictPlot)\n",
    "plt.show()"
   ]
  },
  {
   "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}