LSTM.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import datetime\n",
    "import boto3\n",
    "from keras.preprocessing.sequence import TimeseriesGenerator\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Dropout\n",
    "from keras.layers import LSTM\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from matplotlib import pyplot\n",
    "from pandas.tseries.offsets import DateOffset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#The boto library is used to read the values from the IoT analytics dataset\n",
    "#client = boto3.client(\"iotanalytics\")\n",
    "#region = 'dublin'\n",
    "#device_type = 'evc'\n",
    "#dataset_name = 'grianity_analytics_' + region + '_' + device_type + '_dataset'\n",
    "#dataset_url = client.get_dataset_content(datasetname=dataset_name)\n",
    "#df = pd.read_csv(dataset_url, header=0, parse_dates=[0], index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/eoin/.local/lib/python2.7/site-packages/sklearn/preprocessing/data.py:334: DataConversionWarning: Data with input dtype int64 were all converted to float64 by MinMaxScaler.\n",
      "  return self.partial_fit(X, y)\n"
     ]
    }
   ],
   "source": [
    "df = pd.read_csv(\"ev_data.csv\", header=0, parse_dates=[0], index_col=0)\n",
    "df = df.resample('H').sum()\n",
    "\n",
    "#Using 90% of the data for training\n",
    "num_data = len(df)\n",
    "num_train = int(0.90 * num_data)\n",
    "num_test = num_data - num_train\n",
    "train_data = df[0:num_train]\n",
    "test_data = df[num_test:]\n",
    "\n",
    "\n",
    "# Normalising the data\n",
    "scaler = MinMaxScaler()\n",
    "scaler.fit(train_data)\n",
    "scaler.fit(test_data)\n",
    "train_data = scaler.transform(train_data)\n",
    "test_data = scaler.transform(test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    " n_input = 24\n",
    "n_features = 1\n",
    "generator = TimeseriesGenerator(train_data, train_data, length=n_input, batch_size=5)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/180\n",
      "38/38 [==============================] - 3s 70ms/step - loss: 0.0645\n",
      "Epoch 2/180\n",
      "38/38 [==============================] - 2s 47ms/step - loss: 0.0512\n",
      "Epoch 3/180\n",
      "38/38 [==============================] - 1s 39ms/step - loss: 0.0501\n",
      "Epoch 4/180\n",
      "38/38 [==============================] - 1s 39ms/step - loss: 0.0394\n",
      "Epoch 5/180\n",
      "38/38 [==============================] - 2s 40ms/step - loss: 0.0364\n",
      "Epoch 6/180\n",
      "38/38 [==============================] - 2s 40ms/step - loss: 0.0274\n",
      "Epoch 7/180\n",
      "38/38 [==============================] - 2s 40ms/step - loss: 0.0254\n",
      "Epoch 8/180\n",
      "38/38 [==============================] - 2s 40ms/step - loss: 0.0202\n",
      "Epoch 9/180\n",
      "38/38 [==============================] - 2s 41ms/step - loss: 0.0182\n",
      "Epoch 10/180\n",
      "38/38 [==============================] - 2s 40ms/step - loss: 0.0183\n",
      "Epoch 11/180\n",
      "38/38 [==============================] - 2s 41ms/step - loss: 0.0169\n",
      "Epoch 12/180\n",
      "38/38 [==============================] - 2s 47ms/step - loss: 0.0176\n",
      "Epoch 13/180\n",
      "38/38 [==============================] - 2s 49ms/step - loss: 0.0166\n",
      "Epoch 14/180\n",
      "38/38 [==============================] - 2s 40ms/step - loss: 0.0154\n",
      "Epoch 15/180\n",
      "38/38 [==============================] - 2s 43ms/step - loss: 0.0167\n",
      "Epoch 16/180\n",
      "38/38 [==============================] - 2s 45ms/step - loss: 0.0143\n",
      "Epoch 17/180\n",
      "38/38 [==============================] - 2s 44ms/step - loss: 0.0161\n",
      "Epoch 18/180\n",
      "38/38 [==============================] - 2s 51ms/step - loss: 0.0143\n",
      "Epoch 19/180\n",
      "38/38 [==============================] - 2s 46ms/step - loss: 0.0154\n",
      "Epoch 20/180\n",
      "38/38 [==============================] - 2s 44ms/step - loss: 0.0154\n",
      "Epoch 21/180\n",
      "38/38 [==============================] - 2s 41ms/step - loss: 0.0152\n",
      "Epoch 22/180\n",
      "38/38 [==============================] - 2s 45ms/step - loss: 0.0151\n",
      "Epoch 23/180\n",
      "38/38 [==============================] - 2s 44ms/step - loss: 0.0162\n",
      "Epoch 24/180\n",
      "38/38 [==============================] - 2s 44ms/step - loss: 0.0151\n",
      "Epoch 25/180\n",
      "38/38 [==============================] - 2s 43ms/step - loss: 0.0158\n",
      "Epoch 26/180\n",
      "38/38 [==============================] - 2s 47ms/step - loss: 0.0150\n",
      "Epoch 27/180\n",
      "38/38 [==============================] - 2s 52ms/step - loss: 0.0138\n",
      "Epoch 28/180\n",
      "38/38 [==============================] - 3s 73ms/step - loss: 0.0148\n",
      "Epoch 29/180\n",
      "38/38 [==============================] - 2s 59ms/step - loss: 0.0141\n",
      "Epoch 30/180\n",
      "38/38 [==============================] - 2s 51ms/step - loss: 0.0148\n",
      "Epoch 31/180\n",
      "38/38 [==============================] - 2s 54ms/step - loss: 0.0143\n",
      "Epoch 32/180\n",
      "38/38 [==============================] - 2s 47ms/step - loss: 0.0145\n",
      "Epoch 33/180\n",
      "38/38 [==============================] - 1s 35ms/step - loss: 0.0143\n",
      "Epoch 34/180\n",
      "38/38 [==============================] - 1s 39ms/step - loss: 0.0142\n",
      "Epoch 35/180\n",
      "38/38 [==============================] - 2s 48ms/step - loss: 0.0145\n",
      "Epoch 36/180\n",
      "38/38 [==============================] - 2s 56ms/step - loss: 0.0141\n",
      "Epoch 37/180\n",
      "38/38 [==============================] - 2s 46ms/step - loss: 0.0140\n",
      "Epoch 38/180\n",
      "38/38 [==============================] - 1s 37ms/step - loss: 0.0148: 0s - loss:\n",
      "Epoch 39/180\n",
      "38/38 [==============================] - 2s 57ms/step - loss: 0.0150\n",
      "Epoch 40/180\n",
      "38/38 [==============================] - 2s 61ms/step - loss: 0.0138: 1s -\n",
      "Epoch 41/180\n",
      "38/38 [==============================] - 3s 69ms/step - loss: 0.0147\n",
      "Epoch 42/180\n",
      "38/38 [==============================] - 3s 87ms/step - loss: 0.0132\n",
      "Epoch 43/180\n",
      "38/38 [==============================] - 3s 72ms/step - loss: 0.0133\n",
      "Epoch 44/180\n",
      "38/38 [==============================] - 3s 78ms/step - loss: 0.0151\n",
      "Epoch 45/180\n",
      "38/38 [==============================] - 5s 139ms/step - loss: 0.0138\n",
      "Epoch 46/180\n",
      "38/38 [==============================] - 3s 86ms/step - loss: 0.0140\n",
      "Epoch 47/180\n",
      "38/38 [==============================] - 3s 72ms/step - loss: 0.0137\n",
      "Epoch 48/180\n",
      "38/38 [==============================] - 9s 228ms/step - loss: 0.0136\n",
      "Epoch 49/180\n",
      "38/38 [==============================] - 36s 943ms/step - loss: 0.0128\n",
      "Epoch 50/180\n",
      "16/38 [===========>..................] - ETA: 12s - loss: 0.0136"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/eoin/.local/lib/python2.7/site-packages/keras/callbacks.py:120: UserWarning: Method on_batch_end() is slow compared to the batch update (0.117804). Check your callbacks.\n",
      "  % delta_t_median)\n",
      "/home/eoin/.local/lib/python2.7/site-packages/keras/callbacks.py:120: UserWarning: Method on_batch_end() is slow compared to the batch update (0.111991). Check your callbacks.\n",
      "  % delta_t_median)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "38/38 [==============================] - 11s 290ms/step - loss: 0.0145\n",
      "Epoch 51/180\n",
      "38/38 [==============================] - 4s 117ms/step - loss: 0.0136\n",
      "Epoch 52/180\n",
      "38/38 [==============================] - 4s 95ms/step - loss: 0.0139\n",
      "Epoch 53/180\n",
      "38/38 [==============================] - 2s 56ms/step - loss: 0.0139\n",
      "Epoch 54/180\n",
      "38/38 [==============================] - 1s 36ms/step - loss: 0.0139\n",
      "Epoch 55/180\n",
      "38/38 [==============================] - 2s 46ms/step - loss: 0.0150\n",
      "Epoch 56/180\n",
      "38/38 [==============================] - 1s 36ms/step - loss: 0.0130: 0s - loss: 0.\n",
      "Epoch 57/180\n",
      "38/38 [==============================] - 1s 38ms/step - loss: 0.0137\n",
      "Epoch 58/180\n",
      "38/38 [==============================] - 1s 36ms/step - loss: 0.0145\n",
      "Epoch 59/180\n",
      "38/38 [==============================] - 1s 36ms/step - loss: 0.0144\n",
      "Epoch 60/180\n",
      "38/38 [==============================] - 1s 37ms/step - loss: 0.0133\n",
      "Epoch 61/180\n",
      "38/38 [==============================] - 1s 36ms/step - loss: 0.0146\n",
      "Epoch 62/180\n",
      "38/38 [==============================] - 2s 44ms/step - loss: 0.0125\n",
      "Epoch 63/180\n",
      "38/38 [==============================] - 1s 37ms/step - loss: 0.0130\n",
      "Epoch 64/180\n",
      "38/38 [==============================] - 1s 37ms/step - loss: 0.0137: 0s - loss: 0\n",
      "Epoch 65/180\n",
      "38/38 [==============================] - 1s 35ms/step - loss: 0.0139\n",
      "Epoch 66/180\n",
      "38/38 [==============================] - 1s 36ms/step - loss: 0.0127\n",
      "Epoch 67/180\n",
      "38/38 [==============================] - 1s 36ms/step - loss: 0.0140\n",
      "Epoch 68/180\n",
      "38/38 [==============================] - ETA: 0s - loss: 0.012 - 1s 35ms/step - loss: 0.0131\n",
      "Epoch 69/180\n",
      "38/38 [==============================] - 1s 36ms/step - loss: 0.0131\n",
      "Epoch 70/180\n",
      "38/38 [==============================] - 1s 37ms/step - loss: 0.0126: 0s - los\n",
      "Epoch 71/180\n",
      "38/38 [==============================] - 2s 43ms/step - loss: 0.0130\n",
      "Epoch 72/180\n",
      "38/38 [==============================] - 1s 38ms/step - loss: 0.0142\n",
      "Epoch 73/180\n",
      "38/38 [==============================] - 1s 39ms/step - loss: 0.0125\n",
      "Epoch 74/180\n",
      "38/38 [==============================] - 2s 57ms/step - loss: 0.0130\n",
      "Epoch 75/180\n",
      "38/38 [==============================] - 3s 90ms/step - loss: 0.0136\n",
      "Epoch 76/180\n",
      "38/38 [==============================] - 10s 274ms/step - loss: 0.01400s - loss\n",
      "Epoch 77/180\n",
      "38/38 [==============================] - 23s 597ms/step - loss: 0.0142\n",
      "Epoch 78/180\n",
      "38/38 [==============================] - 13s 340ms/step - loss: 0.0131\n",
      "Epoch 79/180\n",
      "38/38 [==============================] - 3s 85ms/step - loss: 0.0124\n",
      "Epoch 80/180\n",
      "38/38 [==============================] - 2s 53ms/step - loss: 0.0133\n",
      "Epoch 81/180\n",
      "38/38 [==============================] - 1s 35ms/step - loss: 0.0139\n",
      "Epoch 82/180\n",
      "38/38 [==============================] - 1s 35ms/step - loss: 0.0129\n",
      "Epoch 83/180\n",
      "38/38 [==============================] - 1s 39ms/step - loss: 0.0122\n",
      "Epoch 84/180\n",
      "38/38 [==============================] - 2s 41ms/step - loss: 0.0130\n",
      "Epoch 85/180\n",
      "38/38 [==============================] - 1s 34ms/step - loss: 0.0129\n",
      "Epoch 86/180\n",
      "38/38 [==============================] - 1s 37ms/step - loss: 0.0136\n",
      "Epoch 87/180\n",
      "38/38 [==============================] - 1s 35ms/step - loss: 0.0126\n",
      "Epoch 88/180\n",
      "38/38 [==============================] - 1s 35ms/step - loss: 0.0146\n",
      "Epoch 89/180\n",
      "38/38 [==============================] - 1s 35ms/step - loss: 0.0126: 0s - los\n",
      "Epoch 90/180\n",
      "38/38 [==============================] - 1s 34ms/step - loss: 0.0126\n",
      "Epoch 91/180\n",
      "38/38 [==============================] - 2s 50ms/step - loss: 0.0132\n",
      "Epoch 92/180\n",
      "38/38 [==============================] - 195s 5s/step - loss: 0.0128\n",
      "Epoch 93/180\n",
      "38/38 [==============================] - 4s 104ms/step - loss: 0.0136\n",
      "Epoch 94/180\n",
      "38/38 [==============================] - 3s 78ms/step - loss: 0.0130\n",
      "Epoch 95/180\n",
      "38/38 [==============================] - 2s 63ms/step - loss: 0.0120\n",
      "Epoch 96/180\n",
      "38/38 [==============================] - 2s 61ms/step - loss: 0.0130\n",
      "Epoch 97/180\n",
      "38/38 [==============================] - 2s 62ms/step - loss: 0.0125\n",
      "Epoch 98/180\n",
      "38/38 [==============================] - 2s 64ms/step - loss: 0.0122\n",
      "Epoch 99/180\n",
      "38/38 [==============================] - 2s 55ms/step - loss: 0.0126\n",
      "Epoch 100/180\n",
      "38/38 [==============================] - 2s 63ms/step - loss: 0.0128\n",
      "Epoch 101/180\n",
      "38/38 [==============================] - 2s 62ms/step - loss: 0.0115\n",
      "Epoch 102/180\n",
      "38/38 [==============================] - 2s 64ms/step - loss: 0.0120\n",
      "Epoch 103/180\n",
      "38/38 [==============================] - 2s 63ms/step - loss: 0.0125\n",
      "Epoch 104/180\n",
      "38/38 [==============================] - 2s 56ms/step - loss: 0.0123\n",
      "Epoch 105/180\n",
      "38/38 [==============================] - 2s 62ms/step - loss: 0.0120\n",
      "Epoch 106/180\n",
      "38/38 [==============================] - 2s 60ms/step - loss: 0.0123\n",
      "Epoch 107/180\n",
      "38/38 [==============================] - 3s 69ms/step - loss: 0.0147\n",
      "Epoch 108/180\n",
      "38/38 [==============================] - 2s 57ms/step - loss: 0.0126\n",
      "Epoch 109/180\n",
      "38/38 [==============================] - 2s 59ms/step - loss: 0.0124\n",
      "Epoch 110/180\n",
      "38/38 [==============================] - 2s 62ms/step - loss: 0.0116\n",
      "Epoch 111/180\n",
      "38/38 [==============================] - 2s 59ms/step - loss: 0.0127\n",
      "Epoch 112/180\n",
      "38/38 [==============================] - 2s 61ms/step - loss: 0.0122\n",
      "Epoch 113/180\n",
      "38/38 [==============================] - 2s 59ms/step - loss: 0.0121\n",
      "Epoch 114/180\n",
      "38/38 [==============================] - 2s 65ms/step - loss: 0.0123\n",
      "Epoch 115/180\n",
      "38/38 [==============================] - 13s 331ms/step - loss: 0.0116\n",
      "Epoch 116/180\n",
      "38/38 [==============================] - 3s 75ms/step - loss: 0.0128\n",
      "Epoch 117/180\n",
      "38/38 [==============================] - 2s 50ms/step - loss: 0.0112\n",
      "Epoch 118/180\n",
      "38/38 [==============================] - 2s 55ms/step - loss: 0.0123\n",
      "Epoch 119/180\n",
      "38/38 [==============================] - 2s 51ms/step - loss: 0.0128\n",
      "Epoch 120/180\n",
      "38/38 [==============================] - 2s 65ms/step - loss: 0.0128\n",
      "Epoch 121/180\n",
      "38/38 [==============================] - 3s 84ms/step - loss: 0.0129\n",
      "Epoch 122/180\n",
      "38/38 [==============================] - 2s 58ms/step - loss: 0.0118\n",
      "Epoch 123/180\n",
      "38/38 [==============================] - 3s 75ms/step - loss: 0.0121\n",
      "Epoch 124/180\n",
      "38/38 [==============================] - 3s 79ms/step - loss: 0.0122\n",
      "Epoch 125/180\n",
      "38/38 [==============================] - 2s 56ms/step - loss: 0.0122\n",
      "Epoch 126/180\n",
      "38/38 [==============================] - 2s 54ms/step - loss: 0.0125\n",
      "Epoch 127/180\n",
      "38/38 [==============================] - 3s 74ms/step - loss: 0.0118\n",
      "Epoch 128/180\n",
      "38/38 [==============================] - 3s 81ms/step - loss: 0.0112\n",
      "Epoch 129/180\n",
      "38/38 [==============================] - 3s 75ms/step - loss: 0.0123\n",
      "Epoch 130/180\n",
      "38/38 [==============================] - 2s 65ms/step - loss: 0.0128\n",
      "Epoch 131/180\n",
      "38/38 [==============================] - 2s 56ms/step - loss: 0.0135\n",
      "Epoch 132/180\n",
      "38/38 [==============================] - 2s 52ms/step - loss: 0.0116\n",
      "Epoch 133/180\n",
      "38/38 [==============================] - 2s 56ms/step - loss: 0.0116\n",
      "Epoch 134/180\n",
      "38/38 [==============================] - 2s 50ms/step - loss: 0.0119\n",
      "Epoch 135/180\n",
      "38/38 [==============================] - 2s 52ms/step - loss: 0.0118\n",
      "Epoch 136/180\n",
      "38/38 [==============================] - 2s 61ms/step - loss: 0.0132\n",
      "Epoch 137/180\n",
      "38/38 [==============================] - 3s 68ms/step - loss: 0.0115\n",
      "Epoch 138/180\n",
      "38/38 [==============================] - 2s 52ms/step - loss: 0.0120\n",
      "Epoch 139/180\n",
      "38/38 [==============================] - 2s 63ms/step - loss: 0.0114\n",
      "Epoch 140/180\n",
      "38/38 [==============================] - 2s 65ms/step - loss: 0.0101\n",
      "Epoch 141/180\n",
      "38/38 [==============================] - 3s 72ms/step - loss: 0.0116\n",
      "Epoch 142/180\n",
      "38/38 [==============================] - 2s 59ms/step - loss: 0.0121\n",
      "Epoch 143/180\n",
      "38/38 [==============================] - 2s 55ms/step - loss: 0.0109\n",
      "Epoch 144/180\n",
      "38/38 [==============================] - 2s 59ms/step - loss: 0.0110\n",
      "Epoch 145/180\n",
      "38/38 [==============================] - 3s 70ms/step - loss: 0.0124\n",
      "Epoch 146/180\n",
      "38/38 [==============================] - 3s 83ms/step - loss: 0.0151\n",
      "Epoch 147/180\n",
      "38/38 [==============================] - 3s 83ms/step - loss: 0.0129: 0s - loss: 0.0\n",
      "Epoch 148/180\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "38/38 [==============================] - 3s 70ms/step - loss: 0.0127\n",
      "Epoch 149/180\n",
      "38/38 [==============================] - 2s 63ms/step - loss: 0.0122: \n",
      "Epoch 150/180\n",
      "38/38 [==============================] - 2s 53ms/step - loss: 0.0117: 0s - los\n",
      "Epoch 151/180\n",
      "38/38 [==============================] - 2s 49ms/step - loss: 0.0133\n",
      "Epoch 152/180\n",
      "38/38 [==============================] - 2s 56ms/step - loss: 0.0124\n",
      "Epoch 153/180\n",
      "38/38 [==============================] - 3s 70ms/step - loss: 0.0124\n",
      "Epoch 154/180\n",
      "38/38 [==============================] - 2s 44ms/step - loss: 0.0128\n",
      "Epoch 155/180\n",
      "38/38 [==============================] - 2s 47ms/step - loss: 0.0122: \n",
      "Epoch 156/180\n",
      "38/38 [==============================] - 2s 49ms/step - loss: 0.0117\n",
      "Epoch 157/180\n",
      "38/38 [==============================] - 3s 71ms/step - loss: 0.0116\n",
      "Epoch 158/180\n",
      "38/38 [==============================] - 2s 52ms/step - loss: 0.0111\n",
      "Epoch 159/180\n",
      "38/38 [==============================] - 2s 66ms/step - loss: 0.0110\n",
      "Epoch 160/180\n",
      "38/38 [==============================] - 3s 72ms/step - loss: 0.0118\n",
      "Epoch 161/180\n",
      "38/38 [==============================] - 2s 64ms/step - loss: 0.0136\n",
      "Epoch 162/180\n",
      "38/38 [==============================] - 3s 68ms/step - loss: 0.0140\n",
      "Epoch 163/180\n",
      "38/38 [==============================] - 2s 53ms/step - loss: 0.0137\n",
      "Epoch 164/180\n",
      "38/38 [==============================] - 2s 62ms/step - loss: 0.0116\n",
      "Epoch 165/180\n",
      "38/38 [==============================] - 2s 61ms/step - loss: 0.0111\n",
      "Epoch 166/180\n",
      "38/38 [==============================] - 2s 55ms/step - loss: 0.0112\n",
      "Epoch 167/180\n",
      "38/38 [==============================] - 2s 46ms/step - loss: 0.0108\n",
      "Epoch 168/180\n",
      "38/38 [==============================] - 2s 41ms/step - loss: 0.0097\n",
      "Epoch 169/180\n",
      "38/38 [==============================] - 2s 46ms/step - loss: 0.0093: 0s - lo\n",
      "Epoch 170/180\n",
      "38/38 [==============================] - 2s 48ms/step - loss: 0.0119: 1s\n",
      "Epoch 171/180\n",
      "38/38 [==============================] - 1s 38ms/step - loss: 0.0107\n",
      "Epoch 172/180\n",
      "38/38 [==============================] - 2s 51ms/step - loss: 0.0101\n",
      "Epoch 173/180\n",
      "38/38 [==============================] - 1s 37ms/step - loss: 0.0092\n",
      "Epoch 174/180\n",
      "38/38 [==============================] - 1s 39ms/step - loss: 0.0097\n",
      "Epoch 175/180\n",
      "38/38 [==============================] - 2s 63ms/step - loss: 0.0100\n",
      "Epoch 176/180\n",
      "38/38 [==============================] - 2s 40ms/step - loss: 0.0095\n",
      "Epoch 177/180\n",
      "38/38 [==============================] - 1s 38ms/step - loss: 0.0092\n",
      "Epoch 178/180\n",
      "38/38 [==============================] - 2s 47ms/step - loss: 0.0102\n",
      "Epoch 179/180\n",
      "38/38 [==============================] - 2s 40ms/step - loss: 0.0083\n",
      "Epoch 180/180\n",
      "38/38 [==============================] - 2s 42ms/step - loss: 0.0132\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7fd972fb2f90>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# LSDM model\n",
    "model = Sequential()\n",
    "model.add(LSTM(180, activation='relu', input_shape=(n_input, n_features)))\n",
    "model.add(Dropout(0.15))\n",
    "model.add(Dense(1))\n",
    "model.compile(optimizer='adam', loss='mse')\n",
    "model.fit_generator(generator, epochs=180)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "    prediction_list = []\n",
    "    batch = test_data[-n_input:].reshape((1, n_input, n_features))\n",
    "\n",
    "    for i in range(n_input):\n",
    "        prediction_list.append(model.predict(batch)[0])\n",
    "        batch = np.append(batch[:, 1:, :], [[prediction_list[i]]], axis=1)\n",
    "\n",
    "    df_prediction = pd.DataFrame(scaler.inverse_transform(prediction_list), index=df[-n_input:].index, columns=['prediction'])\n",
    "\n",
    "    df_tests = pd.concat([df, df_prediction], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/eoin/.local/lib/python2.7/site-packages/pandas/plotting/_converter.py:129: FutureWarning: Using an implicitly registered datetime converter for a matplotlib plotting method. The converter was registered by pandas on import. Future versions of pandas will require you to explicitly register matplotlib converters.\n",
      "\n",
      "To register the converters:\n",
      "\t>>> from pandas.plotting import register_matplotlib_converters\n",
      "\t>>> register_matplotlib_converters()\n",
      "  warnings.warn(msg, FutureWarning)\n"
     ]
    },
    {
     "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": [
    "    pyplot.plot(df_tests.index, df_tests['power'], label='original')\n",
    "    pyplot.plot(df_tests.index, df_tests['prediction'], color='red', label='prediction')\n",
    "    pyplot.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "43/43 [==============================] - 2s 40ms/step - loss: 0.0062\n",
      "Epoch 2/100\n",
      "43/43 [==============================] - 2s 38ms/step - loss: 0.0089\n",
      "Epoch 3/100\n",
      "43/43 [==============================] - 2s 38ms/step - loss: 0.0066\n",
      "Epoch 4/100\n",
      "43/43 [==============================] - 2s 38ms/step - loss: 0.0060\n",
      "Epoch 5/100\n",
      "43/43 [==============================] - 2s 52ms/step - loss: 0.0046\n",
      "Epoch 6/100\n",
      "43/43 [==============================] - 3s 61ms/step - loss: 0.0072\n",
      "Epoch 7/100\n",
      "43/43 [==============================] - 3s 65ms/step - loss: 0.0048\n",
      "Epoch 8/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0040\n",
      "Epoch 9/100\n",
      "43/43 [==============================] - 2s 52ms/step - loss: 0.0038\n",
      "Epoch 10/100\n",
      "43/43 [==============================] - 2s 44ms/step - loss: 0.0037\n",
      "Epoch 11/100\n",
      "43/43 [==============================] - 2s 46ms/step - loss: 0.0038\n",
      "Epoch 12/100\n",
      "43/43 [==============================] - 2s 47ms/step - loss: 0.0045\n",
      "Epoch 13/100\n",
      "43/43 [==============================] - 2s 52ms/step - loss: 0.0039\n",
      "Epoch 14/100\n",
      "43/43 [==============================] - 2s 55ms/step - loss: 0.0035\n",
      "Epoch 15/100\n",
      "43/43 [==============================] - 2s 49ms/step - loss: 0.0035\n",
      "Epoch 16/100\n",
      "43/43 [==============================] - 2s 45ms/step - loss: 0.0038\n",
      "Epoch 17/100\n",
      "43/43 [==============================] - 2s 53ms/step - loss: 0.0036\n",
      "Epoch 18/100\n",
      "43/43 [==============================] - 2s 50ms/step - loss: 0.0050\n",
      "Epoch 19/100\n",
      "43/43 [==============================] - 2s 45ms/step - loss: 0.0049\n",
      "Epoch 20/100\n",
      "43/43 [==============================] - 2s 57ms/step - loss: 0.0044\n",
      "Epoch 21/100\n",
      "43/43 [==============================] - 4s 85ms/step - loss: 0.0042\n",
      "Epoch 22/100\n",
      "43/43 [==============================] - 4s 85ms/step - loss: 0.0040\n",
      "Epoch 23/100\n",
      "43/43 [==============================] - 3s 62ms/step - loss: 0.0036\n",
      "Epoch 24/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0038\n",
      "Epoch 25/100\n",
      "43/43 [==============================] - 3s 62ms/step - loss: 0.0042\n",
      "Epoch 26/100\n",
      "43/43 [==============================] - 2s 54ms/step - loss: 0.0032\n",
      "Epoch 27/100\n",
      "43/43 [==============================] - 3s 68ms/step - loss: 0.0031\n",
      "Epoch 28/100\n",
      "43/43 [==============================] - 3s 75ms/step - loss: 0.0041: 0s - loss:\n",
      "Epoch 29/100\n",
      "43/43 [==============================] - 4s 92ms/step - loss: 0.0036\n",
      "Epoch 30/100\n",
      "43/43 [==============================] - 3s 72ms/step - loss: 0.0041\n",
      "Epoch 31/100\n",
      "43/43 [==============================] - 3s 67ms/step - loss: 0.0031\n",
      "Epoch 32/100\n",
      "43/43 [==============================] - 3s 73ms/step - loss: 0.0028\n",
      "Epoch 33/100\n",
      "43/43 [==============================] - 2s 53ms/step - loss: 0.0029\n",
      "Epoch 34/100\n",
      "43/43 [==============================] - 2s 58ms/step - loss: 0.0032\n",
      "Epoch 35/100\n",
      "43/43 [==============================] - 3s 64ms/step - loss: 0.0036\n",
      "Epoch 36/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0035\n",
      "Epoch 37/100\n",
      "43/43 [==============================] - 2s 55ms/step - loss: 0.0031\n",
      "Epoch 38/100\n",
      "43/43 [==============================] - 3s 64ms/step - loss: 0.0047\n",
      "Epoch 39/100\n",
      "43/43 [==============================] - 3s 59ms/step - loss: 0.0039\n",
      "Epoch 40/100\n",
      "43/43 [==============================] - 2s 56ms/step - loss: 0.0036\n",
      "Epoch 41/100\n",
      "43/43 [==============================] - 2s 55ms/step - loss: 0.0031\n",
      "Epoch 42/100\n",
      "43/43 [==============================] - 3s 62ms/step - loss: 0.0032\n",
      "Epoch 43/100\n",
      "43/43 [==============================] - 3s 59ms/step - loss: 0.0037\n",
      "Epoch 44/100\n",
      "43/43 [==============================] - 2s 58ms/step - loss: 0.0042\n",
      "Epoch 45/100\n",
      "43/43 [==============================] - 2s 57ms/step - loss: 0.0049\n",
      "Epoch 46/100\n",
      "43/43 [==============================] - 2s 55ms/step - loss: 0.0043\n",
      "Epoch 47/100\n",
      "43/43 [==============================] - 2s 54ms/step - loss: 0.0042\n",
      "Epoch 48/100\n",
      "43/43 [==============================] - 2s 57ms/step - loss: 0.0106\n",
      "Epoch 49/100\n",
      "43/43 [==============================] - 3s 70ms/step - loss: 0.0082\n",
      "Epoch 50/100\n",
      "43/43 [==============================] - 3s 66ms/step - loss: 0.0083\n",
      "Epoch 51/100\n",
      "43/43 [==============================] - 3s 77ms/step - loss: 0.0059\n",
      "Epoch 52/100\n",
      "43/43 [==============================] - 4s 95ms/step - loss: 0.0061\n",
      "Epoch 53/100\n",
      "43/43 [==============================] - 4s 83ms/step - loss: 0.0050\n",
      "Epoch 54/100\n",
      "43/43 [==============================] - 3s 79ms/step - loss: 0.0043\n",
      "Epoch 55/100\n",
      "43/43 [==============================] - 3s 69ms/step - loss: 0.0036\n",
      "Epoch 56/100\n",
      "43/43 [==============================] - 3s 74ms/step - loss: 0.0039\n",
      "Epoch 57/100\n",
      "43/43 [==============================] - 3s 72ms/step - loss: 0.0040\n",
      "Epoch 58/100\n",
      "43/43 [==============================] - 3s 67ms/step - loss: 0.0036\n",
      "Epoch 59/100\n",
      "43/43 [==============================] - 3s 71ms/step - loss: 0.0036\n",
      "Epoch 60/100\n",
      "43/43 [==============================] - 3s 74ms/step - loss: 0.0033\n",
      "Epoch 61/100\n",
      "43/43 [==============================] - 3s 68ms/step - loss: 0.0029\n",
      "Epoch 62/100\n",
      "43/43 [==============================] - 3s 59ms/step - loss: 0.0026\n",
      "Epoch 63/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0031\n",
      "Epoch 64/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0041\n",
      "Epoch 65/100\n",
      "43/43 [==============================] - 3s 62ms/step - loss: 0.0041\n",
      "Epoch 66/100\n",
      "43/43 [==============================] - 3s 75ms/step - loss: 0.0028\n",
      "Epoch 67/100\n",
      "43/43 [==============================] - 3s 74ms/step - loss: 0.0029\n",
      "Epoch 68/100\n",
      "43/43 [==============================] - 3s 71ms/step - loss: 0.0036\n",
      "Epoch 69/100\n",
      "43/43 [==============================] - 3s 78ms/step - loss: 0.0035\n",
      "Epoch 70/100\n",
      "43/43 [==============================] - 3s 70ms/step - loss: 0.0033:\n",
      "Epoch 71/100\n",
      "43/43 [==============================] - 3s 69ms/step - loss: 0.0029\n",
      "Epoch 72/100\n",
      "43/43 [==============================] - 3s 70ms/step - loss: 0.0031\n",
      "Epoch 73/100\n",
      "43/43 [==============================] - 3s 70ms/step - loss: 0.0028\n",
      "Epoch 74/100\n",
      "43/43 [==============================] - 3s 67ms/step - loss: 0.0029\n",
      "Epoch 75/100\n",
      "43/43 [==============================] - 3s 61ms/step - loss: 0.0035\n",
      "Epoch 76/100\n",
      "43/43 [==============================] - 3s 61ms/step - loss: 0.0042\n",
      "Epoch 77/100\n",
      "43/43 [==============================] - 3s 62ms/step - loss: 0.0047\n",
      "Epoch 78/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0068\n",
      "Epoch 79/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0042\n",
      "Epoch 80/100\n",
      "43/43 [==============================] - 3s 62ms/step - loss: 0.0035\n",
      "Epoch 81/100\n",
      "43/43 [==============================] - 3s 62ms/step - loss: 0.0031\n",
      "Epoch 82/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0031\n",
      "Epoch 83/100\n",
      "43/43 [==============================] - 3s 61ms/step - loss: 0.0032\n",
      "Epoch 84/100\n",
      "43/43 [==============================] - 3s 61ms/step - loss: 0.0026\n",
      "Epoch 85/100\n",
      "43/43 [==============================] - 3s 61ms/step - loss: 0.0024\n",
      "Epoch 86/100\n",
      "43/43 [==============================] - 3s 65ms/step - loss: 0.0023\n",
      "Epoch 87/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0024\n",
      "Epoch 88/100\n",
      "43/43 [==============================] - 3s 62ms/step - loss: 0.0024\n",
      "Epoch 89/100\n",
      "43/43 [==============================] - 3s 62ms/step - loss: 0.0022\n",
      "Epoch 90/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0023\n",
      "Epoch 91/100\n",
      "43/43 [==============================] - 3s 62ms/step - loss: 0.0024\n",
      "Epoch 92/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0026\n",
      "Epoch 93/100\n",
      "43/43 [==============================] - 3s 64ms/step - loss: 0.0021\n",
      "Epoch 94/100\n",
      "43/43 [==============================] - 3s 59ms/step - loss: 0.0021\n",
      "Epoch 95/100\n",
      "43/43 [==============================] - 3s 61ms/step - loss: 0.0026\n",
      "Epoch 96/100\n",
      "43/43 [==============================] - 3s 61ms/step - loss: 0.0027\n",
      "Epoch 97/100\n",
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0019\n",
      "Epoch 98/100\n",
      "43/43 [==============================] - 3s 66ms/step - loss: 0.0023\n",
      "Epoch 99/100\n",
      "43/43 [==============================] - 3s 61ms/step - loss: 0.0032\n",
      "Epoch 100/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "43/43 [==============================] - 3s 60ms/step - loss: 0.0045\n"
     ]
    }
   ],
   "source": [
    "# Add future date predictions\n",
    "train = df\n",
    "scaler.fit(train)\n",
    "train = scaler.transform(train)\n",
    "generator = TimeseriesGenerator(train, train, length=n_input, batch_size=5)\n",
    "\n",
    "model.fit_generator(generator, epochs=100)\n",
    "prediction_list = []\n",
    "batch = train[-n_input:].reshape((1, n_input, n_features))\n",
    "\n",
    "for i in range(n_input):\n",
    "    prediction_list.append(model.predict(batch)[0])\n",
    "    batch = np.append(batch[:, 1:, :], [[prediction_list[i]]], axis=1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "add_dates = [df.index[-1] + DateOffset(hours=x) for x in range(0, 25)]\n",
    "future_dates = pd.DataFrame(index=add_dates[1:], columns=df.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_prediction = pd.DataFrame(scaler.inverse_transform(prediction_list), index=future_dates[-n_input:].index, columns=['prediction'])\n",
    "df_proj = pd.concat([df, df_prediction], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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": [
    "\n",
    "pyplot.plot(df_proj.index, df_proj['power'], label='original')\n",
    "pyplot.plot(df_proj.index, df_proj['prediction'], color='red', label='prediction')\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "2.7.17"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}