Battery-SOC-CNN-BiLSTM-Attention.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "dae8b1e5-62db-4768-816e-b55104344c8f",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import seaborn as sns \n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.svm import SVR\n",
    "from mat2json import loadMat\n",
    "from util import getBatteryCapacity, getChargingValues, getDischargingValues, getDataframe, series_to_supervised, rollingAverage\n",
    "from keras.callbacks import EarlyStopping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "ef93297f-9397-444c-b689-5a27c8d8dcdb",
   "metadata": {},
   "outputs": [],
   "source": [
    "import keras.backend as K\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "def get_activations(model, inputs, print_shape_only=False, layer_name=None):\n",
    "    # From: https://github.com/philipperemy/keras-visualize-activations\n",
    "    activations = []\n",
    "    inp = model.input\n",
    "    if layer_name is None:\n",
    "        outputs = [layer.output for layer in model.layers]\n",
    "    else:\n",
    "        outputs = [layer.output for layer in model.layers if layer.name == layer_name]  # all layer outputs\n",
    "    funcs = [K.function([inp] + [K.learning_phase()], [out]) for out in outputs]  # evaluation functions\n",
    "    layer_outputs = [func([inputs, 1.])[0] for func in funcs]\n",
    "    for layer_activations in layer_outputs:\n",
    "        activations.append(layer_activations)\n",
    "        if print_shape_only:\n",
    "            print(layer_activations.shape)\n",
    "        else:\n",
    "            print(layer_activations.shape)\n",
    "            print(layer_activations)\n",
    "    return activations\n",
    "\n",
    "\n",
    "def get_data(n, input_dim, attention_column=1):\n",
    "    \"\"\"\n",
    "    :param n: the number of samples to retrieve.\n",
    "    :param input_dim: the number of dimensions of each element in the series.\n",
    "    :param attention_column: the column linked to the target. Everything else is purely random.\n",
    "    :return: x: model inputs, y: model targets\n",
    "    \"\"\"\n",
    "    x = np.random.standard_normal(size=(n, input_dim))\n",
    "    y = np.random.randint(low=0, high=2, size=(n, 1))\n",
    "    x[:, attention_column] = y[:, 0]\n",
    "    return x, y\n",
    "\n",
    "\n",
    "def get_data_recurrent(n, time_steps, input_dim, attention_column=10):\n",
    "    \"\"\"\n",
    "    :param n: the number of samples to retrieve.\n",
    "    :param time_steps: the number of time steps of your series.\n",
    "    :param input_dim: the number of dimensions of each element in the series.\n",
    "    :param attention_column: the column linked to the target. Everything else is purely random.\n",
    "    :return: x: model inputs, y: model targets\n",
    "    \"\"\"\n",
    "    x = np.random.standard_normal(size=(n, time_steps, input_dim))\n",
    "    y = np.random.randint(low=0, high=2, size=(n, 1))\n",
    "    x[:, attention_column, :] = np.tile(y[:], (1, input_dim))\n",
    "    return x, y\n",
    "\n",
    "\n",
    "def get_data_recurrent2(n, time_steps, input_dim, attention_dim=5):\n",
    "    x = np.random.standard_normal(size=(n, time_steps, input_dim))\n",
    "    y = np.random.randint(low=0, high=2, size=(n, 1))\n",
    "    x[:,:,attention_dim] =  np.tile(y[:], (1, time_steps))\n",
    "\n",
    "\n",
    "    return x,y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "14dfa77d-2834-4024-9a47-1095eaae652c",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "from keras.layers import Input, Dense, LSTM, merge ,Conv1D,Dropout,Bidirectional,Multiply\n",
    "from keras.models import Model\n",
    "\n",
    "#from attention_utils import get_activations\n",
    "from keras.layers import merge\n",
    "from keras.layers import Multiply\n",
    "from keras.layers.core import *\n",
    "from keras.layers.recurrent import LSTM\n",
    "from keras.models import *\n",
    "\n",
    "SINGLE_ATTENTION_VECTOR = False\n",
    "def attention_3d_block(inputs):\n",
    "    # inputs.shape = (batch_size, time_steps, input_dim)\n",
    "    input_dim = int(inputs.shape[2])\n",
    "    a = inputs\n",
    "    a = Dense(input_dim, activation='softmax')(a)\n",
    "    if SINGLE_ATTENTION_VECTOR:\n",
    "        a = Lambda(lambda x: K.mean(x, axis=1), name='dim_reduction')(a)\n",
    "        a = RepeatVector(input_dim)(a)\n",
    "    a_probs = Permute((1, 2), name='attention_vec')(a)\n",
    "\n",
    "    #output_attention_mul = merge([inputs, a_probs], name='attention_mul', mode='mul')\n",
    "    output_attention_mul = Multiply()([inputs, a_probs])\n",
    "    return output_attention_mul\n",
    "\n",
    "# Another way of writing the attention mechanism is suitable for the use of the above error source:https://blog.csdn.net/uhauha2929/article/details/80733255\n",
    "def attention_3d_block2(inputs, single_attention_vector=False):\n",
    "    # If the upper layer is LSTM, you need return_sequences=True\n",
    "    # inputs.shape = (batch_size, time_steps, input_dim)\n",
    "    time_steps = K.int_shape(inputs)[1]\n",
    "    input_dim = K.int_shape(inputs)[2]\n",
    "    a = Permute((2, 1))(inputs)\n",
    "    a = Dense(time_steps, activation='softmax')(a)\n",
    "    if single_attention_vector:\n",
    "        a = Lambda(lambda x: K.mean(x, axis=1))(a)\n",
    "        a = RepeatVector(input_dim)(a)\n",
    "\n",
    "    a_probs = Permute((2, 1))(a)\n",
    "    # Multiplied by the attention weight, but there is no summation, it seems to have little effect\n",
    "    # If you classify tasks, you can do Flatten expansion\n",
    "    # element-wise\n",
    "    output_attention_mul = Multiply()([inputs, a_probs])\n",
    "    return output_attention_mul\n",
    "\n",
    "\n",
    "\n",
    "def create_dataset(dataset, look_back):\n",
    "    '''\n",
    "    Processing the data\n",
    "    '''\n",
    "    dataX, dataY = [], []\n",
    "    for i in range(len(dataset)-look_back-1):\n",
    "        a = dataset[i:(i+look_back),:]\n",
    "        dataX.append(a)\n",
    "        dataY.append(dataset[i + look_back,:])\n",
    "    TrainX = np.array(dataX)\n",
    "    Train_Y = np.array(dataY)\n",
    "\n",
    "    return TrainX, Train_Y\n",
    "\n",
    "# Multidimensional normalization returns data and maximum and minimum values\n",
    "def NormalizeMult(data):\n",
    "    #normalize used for denormalization\n",
    "    data = np.array(data)\n",
    "    normalize = np.arange(2*data.shape[1],dtype='float64')\n",
    "\n",
    "    normalize = normalize.reshape(data.shape[1],2)\n",
    "    print(normalize.shape)\n",
    "    for i in range(0,data.shape[1]):\n",
    "        #Column i\n",
    "        list = data[:,i]\n",
    "        listlow,listhigh =  np.percentile(list, [0, 100])\n",
    "        # print(i)\n",
    "        normalize[i,0] = listlow\n",
    "        normalize[i,1] = listhigh\n",
    "        delta = listhigh - listlow\n",
    "        if delta != 0:\n",
    "            #Row j\n",
    "            for j in range(0,data.shape[0]):\n",
    "                data[j,i]  =  (data[j,i] - listlow)/delta\n",
    "    #np.save(\"./normalize.npy\",normalize)\n",
    "    return  data,normalize\n",
    "\n",
    "# Multidimensional denormalization\n",
    "def FNormalizeMult(data,normalize):\n",
    "    data = np.array(data)\n",
    "    for i in  range(0,data.shape[1]):\n",
    "        listlow =  normalize[i,0]\n",
    "        listhigh = normalize[i,1]\n",
    "        delta = listhigh - listlow\n",
    "        if delta != 0:\n",
    "            #Row j\n",
    "            for j in range(0,data.shape[0]):\n",
    "                data[j,i]  =  data[j,i]*delta + listlow\n",
    "\n",
    "    return data\n",
    "\n",
    "\n",
    "def attention_model():\n",
    "    inputs = Input(shape=(TIME_STEPS, INPUT_DIMS))\n",
    "\n",
    "    x = Conv1D(filters = 64, kernel_size = 1, activation = 'relu')(inputs)  #, padding = 'same'\n",
    "    x = Dropout(0.3)(x)\n",
    "\n",
    "    #lstm_out = Bidirectional(LSTM(lstm_units, activation='relu'), name='bilstm')(x)\n",
    "    #For GPU you can use CuDNNLSTM\n",
    "    lstm_out = Bidirectional(LSTM(lstm_units, return_sequences=True))(x)\n",
    "    lstm_out = Dropout(0.3)(lstm_out)\n",
    "    attention_mul = attention_3d_block(lstm_out)\n",
    "    attention_mul = Flatten()(attention_mul)\n",
    "\n",
    "    #output = Dense(1, activation='sigmoid')(attention_mul)\n",
    "    output = Dense(1, activation='linear')(attention_mul)\n",
    "    model = Model(inputs=[inputs], outputs=output)\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "75a3bfa7-cdea-48c3-8259-ed8cbf03e73b",
   "metadata": {},
   "outputs": [],
   "source": [
    "B0005 = loadMat('B0005.mat')\n",
    "B0006 = loadMat('B0006.mat')\n",
    "B0007 = loadMat('B0007.mat')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "f0ea6226-777b-47c8-b0ef-0f825bc4dfdf",
   "metadata": {},
   "outputs": [],
   "source": [
    "dfB0005 = getDataframe(B0005)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "e75c3427-7e4d-4db0-bb4b-a3be708f28e2",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/aimran/.local/lib/python3.6/site-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.\n",
      "  FutureWarning\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='cycle', ylabel='capacity'>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.residplot(dfB0005['cycle'], dfB0005['capacity'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "141fa1fb-933d-4c5e-a044-d8930d59664c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>cycle</th>\n",
       "      <th>capacity</th>\n",
       "      <th>max_discharge_temp</th>\n",
       "      <th>max_charge_temp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>2.035338</td>\n",
       "      <td>39.162987</td>\n",
       "      <td>27.355870</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>2.025140</td>\n",
       "      <td>39.246203</td>\n",
       "      <td>31.874973</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>2.013326</td>\n",
       "      <td>38.999202</td>\n",
       "      <td>32.149173</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>2.013285</td>\n",
       "      <td>38.843628</td>\n",
       "      <td>32.017074</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>2.000528</td>\n",
       "      <td>38.977989</td>\n",
       "      <td>31.921070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>163</th>\n",
       "      <td>164</td>\n",
       "      <td>1.153818</td>\n",
       "      <td>41.149193</td>\n",
       "      <td>28.612333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>164</th>\n",
       "      <td>165</td>\n",
       "      <td>1.164401</td>\n",
       "      <td>41.262683</td>\n",
       "      <td>28.542199</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>165</th>\n",
       "      <td>166</td>\n",
       "      <td>1.158797</td>\n",
       "      <td>41.457918</td>\n",
       "      <td>28.565426</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>166</th>\n",
       "      <td>167</td>\n",
       "      <td>1.174975</td>\n",
       "      <td>41.116662</td>\n",
       "      <td>28.631428</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>167</th>\n",
       "      <td>168</td>\n",
       "      <td>1.185675</td>\n",
       "      <td>41.362638</td>\n",
       "      <td>28.658802</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>168 rows × 4 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     cycle  capacity  max_discharge_temp  max_charge_temp\n",
       "0        1  2.035338           39.162987        27.355870\n",
       "1        2  2.025140           39.246203        31.874973\n",
       "2        3  2.013326           38.999202        32.149173\n",
       "3        4  2.013285           38.843628        32.017074\n",
       "4        5  2.000528           38.977989        31.921070\n",
       "..     ...       ...                 ...              ...\n",
       "163    164  1.153818           41.149193        28.612333\n",
       "164    165  1.164401           41.262683        28.542199\n",
       "165    166  1.158797           41.457918        28.565426\n",
       "166    167  1.174975           41.116662        28.631428\n",
       "167    168  1.185675           41.362638        28.658802\n",
       "\n",
       "[168 rows x 4 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dfB0006 = getDataframe(B0006)\n",
    "dfB0007 = getDataframe(B0007)\n",
    "\n",
    "dfB0006\n",
    "\n",
    "#X = dfB0007['cycle']\n",
    "#Y = dfB0007['capacity']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "8ad7d655-f9d4-437d-b8d1-a5b02e88cbc4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['cycle', 'capacity', 'max_discharge_temp', 'max_charge_temp'], dtype='object')\n",
      "(168, 4)\n"
     ]
    }
   ],
   "source": [
    "dfB0007\n",
    "print(dfB0007.columns)\n",
    "print(dfB0007.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "d334db72-f52b-4f07-a30e-24225156a3fc",
   "metadata": {},
   "outputs": [],
   "source": [
    "INPUT_DIMS = 4\n",
    "TIME_STEPS = 10\n",
    "lstm_units = 64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "af085f94-6faa-490c-b485-a4c27ea98dbe",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4, 2)\n"
     ]
    }
   ],
   "source": [
    "#Normalized\n",
    "data,normalize = NormalizeMult(dfB0007)\n",
    "capacity_data = data[:,0].reshape(len(dfB0007),1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "23a5b86c-406c-4f16-a255-713a30d22799",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_X, _ = create_dataset(data,TIME_STEPS)\n",
    "_ , train_Y = create_dataset(capacity_data,TIME_STEPS)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "c63caadc-d0c9-4b1d-8d46-c1bbd7b9b2b6",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(157, 10, 4)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "92bc2b99-5efb-4d22-bca3-d3e802691058",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(157, 1)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_Y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "c382a988-d52f-4a91-a774-3f69f75a2184",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(157, 10, 4) (157, 1)\n",
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_1 (InputLayer)            [(None, 10, 4)]      0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv1d (Conv1D)                 (None, 10, 64)       320         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dropout (Dropout)               (None, 10, 64)       0           conv1d[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "bidirectional (Bidirectional)   (None, 10, 128)      66048       dropout[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dropout_1 (Dropout)             (None, 10, 128)      0           bidirectional[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "dense (Dense)                   (None, 10, 128)      16512       dropout_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "attention_vec (Permute)         (None, 10, 128)      0           dense[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "multiply (Multiply)             (None, 10, 128)      0           dropout_1[0][0]                  \n",
      "                                                                 attention_vec[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "flatten (Flatten)               (None, 1280)         0           multiply[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 1)            1281        flatten[0][0]                    \n",
      "==================================================================================================\n",
      "Total params: 84,161\n",
      "Trainable params: 84,161\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n",
      "Epoch 1/100\n",
      "4/4 [==============================] - 5s 164ms/step - loss: 0.2300 - val_loss: 0.8017\n",
      "Epoch 2/100\n",
      "4/4 [==============================] - 0s 20ms/step - loss: 0.2228 - val_loss: 0.7853\n",
      "Epoch 3/100\n",
      "4/4 [==============================] - 0s 19ms/step - loss: 0.2140 - val_loss: 0.7626\n",
      "Epoch 4/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.2011 - val_loss: 0.7265\n",
      "Epoch 5/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.1806 - val_loss: 0.6656\n",
      "Epoch 6/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.1499 - val_loss: 0.5705\n",
      "Epoch 7/100\n",
      "4/4 [==============================] - 0s 20ms/step - loss: 0.1101 - val_loss: 0.4496\n",
      "Epoch 8/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0699 - val_loss: 0.3227\n",
      "Epoch 9/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0466 - val_loss: 0.2116\n",
      "Epoch 10/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0446 - val_loss: 0.1505\n",
      "Epoch 11/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0531 - val_loss: 0.1510\n",
      "Epoch 12/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0474 - val_loss: 0.1749\n",
      "Epoch 13/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0441 - val_loss: 0.2063\n",
      "Epoch 14/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0430 - val_loss: 0.2271\n",
      "Epoch 15/100\n",
      "4/4 [==============================] - 0s 20ms/step - loss: 0.0405 - val_loss: 0.2296\n",
      "Epoch 16/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0412 - val_loss: 0.2190\n",
      "Epoch 17/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0395 - val_loss: 0.2008\n",
      "Epoch 18/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0374 - val_loss: 0.1839\n",
      "Epoch 19/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0343 - val_loss: 0.1745\n",
      "Epoch 20/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0324 - val_loss: 0.1627\n",
      "Epoch 21/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0295 - val_loss: 0.1509\n",
      "Epoch 22/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0265 - val_loss: 0.1404\n",
      "Epoch 23/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0216 - val_loss: 0.1294\n",
      "Epoch 24/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0168 - val_loss: 0.1109\n",
      "Epoch 25/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0096 - val_loss: 0.0780\n",
      "Epoch 26/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0056 - val_loss: 0.0542\n",
      "Epoch 27/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0058 - val_loss: 0.0443\n",
      "Epoch 28/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0050 - val_loss: 0.0366\n",
      "Epoch 29/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0063 - val_loss: 0.0361\n",
      "Epoch 30/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0053 - val_loss: 0.0459\n",
      "Epoch 31/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0047 - val_loss: 0.0386\n",
      "Epoch 32/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0041 - val_loss: 0.0435\n",
      "Epoch 33/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0046 - val_loss: 0.0471\n",
      "Epoch 34/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0040 - val_loss: 0.0426\n",
      "Epoch 35/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0036 - val_loss: 0.0411\n",
      "Epoch 36/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0034 - val_loss: 0.0411\n",
      "Epoch 37/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0036 - val_loss: 0.0415\n",
      "Epoch 38/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0040 - val_loss: 0.0370\n",
      "Epoch 39/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0039 - val_loss: 0.0412\n",
      "Epoch 40/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0028 - val_loss: 0.0407\n",
      "Epoch 41/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0027 - val_loss: 0.0350\n",
      "Epoch 42/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0032 - val_loss: 0.0381\n",
      "Epoch 43/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0029 - val_loss: 0.0405\n",
      "Epoch 44/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0033 - val_loss: 0.0376\n",
      "Epoch 45/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0023 - val_loss: 0.0368\n",
      "Epoch 46/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0025 - val_loss: 0.0360\n",
      "Epoch 47/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0024 - val_loss: 0.0427\n",
      "Epoch 48/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0024 - val_loss: 0.0376\n",
      "Epoch 49/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0020 - val_loss: 0.0337\n",
      "Epoch 50/100\n",
      "4/4 [==============================] - 0s 20ms/step - loss: 0.0020 - val_loss: 0.0350\n",
      "Epoch 51/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0018 - val_loss: 0.0355\n",
      "Epoch 52/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0021 - val_loss: 0.0346\n",
      "Epoch 53/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0021 - val_loss: 0.0364\n",
      "Epoch 54/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0015 - val_loss: 0.0343\n",
      "Epoch 55/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0018 - val_loss: 0.0295\n",
      "Epoch 56/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0016 - val_loss: 0.0339\n",
      "Epoch 57/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0022 - val_loss: 0.0331\n",
      "Epoch 58/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0020 - val_loss: 0.0284\n",
      "Epoch 59/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0017 - val_loss: 0.0320\n",
      "Epoch 60/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0018 - val_loss: 0.0296\n",
      "Epoch 61/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0018 - val_loss: 0.0279\n",
      "Epoch 62/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0019 - val_loss: 0.0324\n",
      "Epoch 63/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0014 - val_loss: 0.0302\n",
      "Epoch 64/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0013 - val_loss: 0.0293\n",
      "Epoch 65/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0016 - val_loss: 0.0290\n",
      "Epoch 66/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0015 - val_loss: 0.0298\n",
      "Epoch 67/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0019 - val_loss: 0.0284\n",
      "Epoch 68/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0022 - val_loss: 0.0307\n",
      "Epoch 69/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - val_loss: 0.0308\n",
      "Epoch 70/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0017 - val_loss: 0.0286\n",
      "Epoch 71/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0013 - val_loss: 0.0305\n",
      "Epoch 72/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0015 - val_loss: 0.0294\n",
      "Epoch 73/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0015 - val_loss: 0.0259\n",
      "Epoch 74/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0016 - val_loss: 0.0284\n",
      "Epoch 75/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0015 - val_loss: 0.0288\n",
      "Epoch 76/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 9.7973e-04 - val_loss: 0.0248\n",
      "Epoch 77/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0014 - val_loss: 0.0255\n",
      "Epoch 78/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0017 - val_loss: 0.0281\n",
      "Epoch 79/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0010 - val_loss: 0.0244\n",
      "Epoch 80/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0016 - val_loss: 0.0255\n",
      "Epoch 81/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0011 - val_loss: 0.0250\n",
      "Epoch 82/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - val_loss: 0.0219\n",
      "Epoch 83/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0015 - val_loss: 0.0235\n",
      "Epoch 84/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0016 - val_loss: 0.0253\n",
      "Epoch 85/100\n",
      "4/4 [==============================] - 0s 20ms/step - loss: 0.0017 - val_loss: 0.0211\n",
      "Epoch 86/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0014 - val_loss: 0.0207\n",
      "Epoch 87/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - val_loss: 0.0238\n",
      "Epoch 88/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0014 - val_loss: 0.0221\n",
      "Epoch 89/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0011 - val_loss: 0.0205\n",
      "Epoch 90/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 8.4634e-04 - val_loss: 0.0225\n",
      "Epoch 91/100\n",
      "4/4 [==============================] - 0s 25ms/step - loss: 0.0010 - val_loss: 0.0213\n",
      "Epoch 92/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 8.4273e-04 - val_loss: 0.0202\n",
      "Epoch 93/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 8.6996e-04 - val_loss: 0.0195\n",
      "Epoch 94/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0010 - val_loss: 0.0191\n",
      "Epoch 95/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - val_loss: 0.0212\n",
      "Epoch 96/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0012 - val_loss: 0.0195\n",
      "Epoch 97/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0011 - val_loss: 0.0189\n",
      "Epoch 98/100\n",
      "4/4 [==============================] - 0s 25ms/step - loss: 7.6174e-04 - val_loss: 0.0187\n",
      "Epoch 99/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0012 - val_loss: 0.0202\n",
      "Epoch 100/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 8.7996e-04 - val_loss: 0.0175\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f3ccc0afc50>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_X, _ = create_dataset(data,TIME_STEPS)\n",
    "_ , train_Y = create_dataset(capacity_data,TIME_STEPS)\n",
    "\n",
    "print(train_X.shape,train_Y.shape)\n",
    "\n",
    "m = attention_model()\n",
    "m.summary()\n",
    "m.compile(optimizer='adam', loss='mse')\n",
    "m.fit([train_X], train_Y, epochs=100, batch_size=32, validation_split=0.2)\n",
    "#m.save(\"./model.h5\")\n",
    "#np.save(\"normalize.npy\",normalize)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "4d7264ed-c623-4bba-bb42-3242bed738cf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "len(train): 134\n",
      "len(test): 34\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "train_size = int(len(dfB0007)*0.8)\n",
    "train = dfB0007.iloc[:train_size,:]\n",
    "test = dfB0007.iloc[train_size:,:]\n",
    "#train, test = train_test_split(df, test_size=0.1)\n",
    "print(\"len(train):\",len(train))\n",
    "print(\"len(test):\",len(test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "77c17a4a-96f1-4bc5-a888-e7f4bed568c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "import keras.backend as K\n",
    "def r2_keras(y_true, y_pred):\n",
    "    \"\"\"Coefficient of Determination \n",
    "    \"\"\"\n",
    "    SS_res =  K.sum(K.square( y_true - y_pred ))\n",
    "    SS_tot = K.sum(K.square( y_true - K.mean(y_true) ) )\n",
    "    return ( 1 - SS_res/(SS_tot + K.epsilon()) )\n",
    "     "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "a1dd0e0b-733a-41e9-af82-54705f724e45",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "rm: cannot remove 'model.h5': No such file or directory\n"
     ]
    }
   ],
   "source": [
    "%rm model.h5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "6f572017-c4d6-4c18-a17e-96d713f427f0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4, 2)\n",
      "(123, 10, 4) (123, 1)\n",
      "Model: \"model_1\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_2 (InputLayer)            [(None, 10, 4)]      0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_1 (Conv1D)               (None, 10, 64)       320         input_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dropout_2 (Dropout)             (None, 10, 64)       0           conv1d_1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "bidirectional_1 (Bidirectional) (None, 10, 128)      66048       dropout_2[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_3 (Dropout)             (None, 10, 128)      0           bidirectional_1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 10, 128)      16512       dropout_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "attention_vec (Permute)         (None, 10, 128)      0           dense_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "multiply_1 (Multiply)           (None, 10, 128)      0           dropout_3[0][0]                  \n",
      "                                                                 attention_vec[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "flatten_1 (Flatten)             (None, 1280)         0           multiply_1[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "dense_3 (Dense)                 (None, 1)            1281        flatten_1[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 84,161\n",
      "Trainable params: 84,161\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n",
      "Epoch 1/100\n",
      "4/4 [==============================] - 5s 171ms/step - loss: 0.2366 - mae: 0.4375 - r2_keras: -330.8606 - val_loss: 0.7970 - val_mae: 0.8911 - val_r2_keras: -270.1305\n",
      "Epoch 2/100\n",
      "4/4 [==============================] - 0s 19ms/step - loss: 0.2290 - mae: 0.4290 - r2_keras: -5.1507 - val_loss: 0.7747 - val_mae: 0.8785 - val_r2_keras: -262.5176\n",
      "Epoch 3/100\n",
      "4/4 [==============================] - 0s 20ms/step - loss: 0.2191 - mae: 0.4180 - r2_keras: -6.5578 - val_loss: 0.7419 - val_mae: 0.8596 - val_r2_keras: -251.3680\n",
      "Epoch 4/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.2056 - mae: 0.4022 - r2_keras: -6.0332 - val_loss: 0.6889 - val_mae: 0.8283 - val_r2_keras: -233.3560\n",
      "Epoch 5/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.1830 - mae: 0.3752 - r2_keras: -2.9531 - val_loss: 0.6103 - val_mae: 0.7793 - val_r2_keras: -206.5904\n",
      "Epoch 6/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.1503 - mae: 0.3316 - r2_keras: -2.4767 - val_loss: 0.5091 - val_mae: 0.7115 - val_r2_keras: -172.1958\n",
      "Epoch 7/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.1075 - mae: 0.2749 - r2_keras: -8.3093 - val_loss: 0.3968 - val_mae: 0.6277 - val_r2_keras: -133.9929\n",
      "Epoch 8/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0710 - mae: 0.2211 - r2_keras: -1.6866 - val_loss: 0.2792 - val_mae: 0.5257 - val_r2_keras: -93.9737\n",
      "Epoch 9/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0456 - mae: 0.1806 - r2_keras: -0.1889 - val_loss: 0.1777 - val_mae: 0.4181 - val_r2_keras: -59.4527\n",
      "Epoch 10/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0408 - mae: 0.1695 - r2_keras: -0.5714 - val_loss: 0.1287 - val_mae: 0.3548 - val_r2_keras: -42.7874\n",
      "Epoch 11/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0486 - mae: 0.1825 - r2_keras: -1.5348 - val_loss: 0.1239 - val_mae: 0.3479 - val_r2_keras: -41.1377\n",
      "Epoch 12/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0473 - mae: 0.1791 - r2_keras: -0.3140 - val_loss: 0.1371 - val_mae: 0.3664 - val_r2_keras: -45.6245\n",
      "Epoch 13/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0394 - mae: 0.1685 - r2_keras: 0.0443 - val_loss: 0.1514 - val_mae: 0.3854 - val_r2_keras: -50.4886\n",
      "Epoch 14/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0378 - mae: 0.1616 - r2_keras: 0.1467 - val_loss: 0.1627 - val_mae: 0.3999 - val_r2_keras: -54.3601\n",
      "Epoch 15/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0387 - mae: 0.1679 - r2_keras: 0.1627 - val_loss: 0.1662 - val_mae: 0.4043 - val_r2_keras: -55.5524\n",
      "Epoch 16/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0375 - mae: 0.1647 - r2_keras: -0.4931 - val_loss: 0.1688 - val_mae: 0.4075 - val_r2_keras: -56.4215\n",
      "Epoch 17/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0327 - mae: 0.1527 - r2_keras: 0.1302 - val_loss: 0.1729 - val_mae: 0.4126 - val_r2_keras: -57.8256\n",
      "Epoch 18/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0343 - mae: 0.1554 - r2_keras: 0.1939 - val_loss: 0.1643 - val_mae: 0.4021 - val_r2_keras: -54.8999\n",
      "Epoch 19/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0295 - mae: 0.1459 - r2_keras: 0.3949 - val_loss: 0.1415 - val_mae: 0.3727 - val_r2_keras: -47.1448\n",
      "Epoch 20/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0264 - mae: 0.1328 - r2_keras: -0.0974 - val_loss: 0.1172 - val_mae: 0.3384 - val_r2_keras: -38.8519\n",
      "Epoch 21/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0253 - mae: 0.1332 - r2_keras: 0.2748 - val_loss: 0.0977 - val_mae: 0.3083 - val_r2_keras: -32.2208\n",
      "Epoch 22/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0206 - mae: 0.1151 - r2_keras: 0.4135 - val_loss: 0.0901 - val_mae: 0.2960 - val_r2_keras: -29.6641\n",
      "Epoch 23/100\n",
      "4/4 [==============================] - 0s 20ms/step - loss: 0.0155 - mae: 0.0954 - r2_keras: 0.6287 - val_loss: 0.0867 - val_mae: 0.2903 - val_r2_keras: -28.4943\n",
      "Epoch 24/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0113 - mae: 0.0858 - r2_keras: 0.7753 - val_loss: 0.0714 - val_mae: 0.2629 - val_r2_keras: -23.2950\n",
      "Epoch 25/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0080 - mae: 0.0769 - r2_keras: 0.8488 - val_loss: 0.0494 - val_mae: 0.2171 - val_r2_keras: -15.8076\n",
      "Epoch 26/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0043 - mae: 0.0561 - r2_keras: -0.7559 - val_loss: 0.0302 - val_mae: 0.1670 - val_r2_keras: -9.2664\n",
      "Epoch 27/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0040 - mae: 0.0535 - r2_keras: 0.2087 - val_loss: 0.0189 - val_mae: 0.1288 - val_r2_keras: -5.4207\n",
      "Epoch 28/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0058 - mae: 0.0650 - r2_keras: -2.4605 - val_loss: 0.0227 - val_mae: 0.1429 - val_r2_keras: -6.7215\n",
      "Epoch 29/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0047 - mae: 0.0564 - r2_keras: 0.9187 - val_loss: 0.0305 - val_mae: 0.1681 - val_r2_keras: -9.3844\n",
      "Epoch 30/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0044 - mae: 0.0563 - r2_keras: 0.8185 - val_loss: 0.0326 - val_mae: 0.1741 - val_r2_keras: -10.0872\n",
      "Epoch 31/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0039 - mae: 0.0511 - r2_keras: 0.5616 - val_loss: 0.0315 - val_mae: 0.1709 - val_r2_keras: -9.7118\n",
      "Epoch 32/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0038 - mae: 0.0486 - r2_keras: 0.6263 - val_loss: 0.0309 - val_mae: 0.1690 - val_r2_keras: -9.4982\n",
      "Epoch 33/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0032 - mae: 0.0457 - r2_keras: 0.8491 - val_loss: 0.0318 - val_mae: 0.1718 - val_r2_keras: -9.8112\n",
      "Epoch 34/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0038 - mae: 0.0496 - r2_keras: 0.8681 - val_loss: 0.0310 - val_mae: 0.1697 - val_r2_keras: -9.5560\n",
      "Epoch 35/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0032 - mae: 0.0438 - r2_keras: 0.6676 - val_loss: 0.0318 - val_mae: 0.1720 - val_r2_keras: -9.8151\n",
      "Epoch 36/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0026 - mae: 0.0409 - r2_keras: 0.9395 - val_loss: 0.0290 - val_mae: 0.1636 - val_r2_keras: -8.8612\n",
      "Epoch 37/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0019 - mae: 0.0356 - r2_keras: 0.9590 - val_loss: 0.0268 - val_mae: 0.1568 - val_r2_keras: -8.1190\n",
      "Epoch 38/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0021 - mae: 0.0357 - r2_keras: 0.9263 - val_loss: 0.0276 - val_mae: 0.1594 - val_r2_keras: -8.3783\n",
      "Epoch 39/100\n",
      "4/4 [==============================] - 0s 20ms/step - loss: 0.0028 - mae: 0.0415 - r2_keras: 0.9394 - val_loss: 0.0254 - val_mae: 0.1524 - val_r2_keras: -7.6346\n",
      "Epoch 40/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0021 - mae: 0.0359 - r2_keras: 0.9652 - val_loss: 0.0235 - val_mae: 0.1461 - val_r2_keras: -6.9857\n",
      "Epoch 41/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0022 - mae: 0.0373 - r2_keras: 0.9539 - val_loss: 0.0215 - val_mae: 0.1390 - val_r2_keras: -6.2985\n",
      "Epoch 42/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0026 - mae: 0.0376 - r2_keras: 0.6743 - val_loss: 0.0195 - val_mae: 0.1317 - val_r2_keras: -5.6344\n",
      "Epoch 43/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0024 - mae: 0.0363 - r2_keras: 0.8907 - val_loss: 0.0192 - val_mae: 0.1304 - val_r2_keras: -5.5164\n",
      "Epoch 44/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0023 - mae: 0.0377 - r2_keras: 0.9414 - val_loss: 0.0211 - val_mae: 0.1377 - val_r2_keras: -6.1701\n",
      "Epoch 45/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0020 - mae: 0.0330 - r2_keras: 0.9590 - val_loss: 0.0225 - val_mae: 0.1429 - val_r2_keras: -6.6514\n",
      "Epoch 46/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0017 - mae: 0.0303 - r2_keras: 0.9657 - val_loss: 0.0226 - val_mae: 0.1433 - val_r2_keras: -6.6823\n",
      "Epoch 47/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0018 - mae: 0.0322 - r2_keras: 0.9630 - val_loss: 0.0204 - val_mae: 0.1354 - val_r2_keras: -5.9486\n",
      "Epoch 48/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0015 - mae: 0.0306 - r2_keras: 0.9663 - val_loss: 0.0199 - val_mae: 0.1335 - val_r2_keras: -5.7681\n",
      "Epoch 49/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0015 - mae: 0.0281 - r2_keras: 0.9689 - val_loss: 0.0211 - val_mae: 0.1381 - val_r2_keras: -6.1838\n",
      "Epoch 50/100\n",
      "4/4 [==============================] - 0s 26ms/step - loss: 0.0012 - mae: 0.0255 - r2_keras: 0.8109 - val_loss: 0.0218 - val_mae: 0.1405 - val_r2_keras: -6.4074\n",
      "Epoch 51/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0019 - mae: 0.0332 - r2_keras: 0.9646 - val_loss: 0.0207 - val_mae: 0.1366 - val_r2_keras: -6.0429\n",
      "Epoch 52/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0015 - mae: 0.0309 - r2_keras: 0.9323 - val_loss: 0.0204 - val_mae: 0.1358 - val_r2_keras: -5.9565\n",
      "Epoch 53/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0015 - mae: 0.0287 - r2_keras: 0.9602 - val_loss: 0.0223 - val_mae: 0.1427 - val_r2_keras: -6.5819\n",
      "Epoch 54/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0014 - mae: 0.0265 - r2_keras: 0.9565 - val_loss: 0.0217 - val_mae: 0.1404 - val_r2_keras: -6.3652\n",
      "Epoch 55/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - mae: 0.0281 - r2_keras: 0.9743 - val_loss: 0.0185 - val_mae: 0.1286 - val_r2_keras: -5.3027\n",
      "Epoch 56/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0013 - mae: 0.0276 - r2_keras: 0.9752 - val_loss: 0.0171 - val_mae: 0.1227 - val_r2_keras: -4.8111\n",
      "Epoch 57/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0014 - mae: 0.0274 - r2_keras: 0.9666 - val_loss: 0.0164 - val_mae: 0.1201 - val_r2_keras: -4.5869\n",
      "Epoch 58/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0014 - mae: 0.0271 - r2_keras: 0.8425 - val_loss: 0.0168 - val_mae: 0.1216 - val_r2_keras: -4.7080\n",
      "Epoch 59/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0012 - mae: 0.0246 - r2_keras: 0.9782 - val_loss: 0.0178 - val_mae: 0.1257 - val_r2_keras: -5.0455\n",
      "Epoch 60/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0012 - mae: 0.0268 - r2_keras: 0.9540 - val_loss: 0.0195 - val_mae: 0.1326 - val_r2_keras: -5.6421\n",
      "Epoch 61/100\n",
      "4/4 [==============================] - 0s 19ms/step - loss: 0.0017 - mae: 0.0282 - r2_keras: 0.9422 - val_loss: 0.0213 - val_mae: 0.1393 - val_r2_keras: -6.2515\n",
      "Epoch 62/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0016 - mae: 0.0286 - r2_keras: 0.9610 - val_loss: 0.0203 - val_mae: 0.1354 - val_r2_keras: -5.8956\n",
      "Epoch 63/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - mae: 0.0272 - r2_keras: 0.9723 - val_loss: 0.0201 - val_mae: 0.1349 - val_r2_keras: -5.8532\n",
      "Epoch 64/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0011 - mae: 0.0242 - r2_keras: 0.9608 - val_loss: 0.0204 - val_mae: 0.1358 - val_r2_keras: -5.9357\n",
      "Epoch 65/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0250 - r2_keras: 0.9786 - val_loss: 0.0174 - val_mae: 0.1241 - val_r2_keras: -4.9235\n",
      "Epoch 66/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - mae: 0.0272 - r2_keras: 0.8339 - val_loss: 0.0185 - val_mae: 0.1285 - val_r2_keras: -5.2918\n",
      "Epoch 67/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0012 - mae: 0.0263 - r2_keras: -8.7389 - val_loss: 0.0215 - val_mae: 0.1401 - val_r2_keras: -6.3264\n",
      "Epoch 68/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0252 - r2_keras: 0.9790 - val_loss: 0.0179 - val_mae: 0.1261 - val_r2_keras: -5.0777\n",
      "Epoch 69/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0245 - r2_keras: 0.9614 - val_loss: 0.0161 - val_mae: 0.1187 - val_r2_keras: -4.4729\n",
      "Epoch 70/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0011 - mae: 0.0247 - r2_keras: 0.9668 - val_loss: 0.0173 - val_mae: 0.1238 - val_r2_keras: -4.8769\n",
      "Epoch 71/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - mae: 0.0259 - r2_keras: 0.9758 - val_loss: 0.0167 - val_mae: 0.1215 - val_r2_keras: -4.6915\n",
      "Epoch 72/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0010 - mae: 0.0236 - r2_keras: 0.9797 - val_loss: 0.0152 - val_mae: 0.1152 - val_r2_keras: -4.1836\n",
      "Epoch 73/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - mae: 0.0269 - r2_keras: -0.3578 - val_loss: 0.0168 - val_mae: 0.1217 - val_r2_keras: -4.7024\n",
      "Epoch 74/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0016 - mae: 0.0280 - r2_keras: 0.9378 - val_loss: 0.0227 - val_mae: 0.1443 - val_r2_keras: -6.7089\n",
      "Epoch 75/100\n",
      "4/4 [==============================] - 0s 20ms/step - loss: 0.0018 - mae: 0.0305 - r2_keras: 0.9673 - val_loss: 0.0205 - val_mae: 0.1365 - val_r2_keras: -5.9805\n",
      "Epoch 76/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 0.0010 - mae: 0.0245 - r2_keras: 0.9712 - val_loss: 0.0173 - val_mae: 0.1237 - val_r2_keras: -4.8684\n",
      "Epoch 77/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0244 - r2_keras: 0.9587 - val_loss: 0.0152 - val_mae: 0.1152 - val_r2_keras: -4.1822\n",
      "Epoch 78/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0012 - mae: 0.0266 - r2_keras: 0.1005 - val_loss: 0.0152 - val_mae: 0.1152 - val_r2_keras: -4.1788\n",
      "Epoch 79/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0245 - r2_keras: 0.9796 - val_loss: 0.0170 - val_mae: 0.1226 - val_r2_keras: -4.7703\n",
      "Epoch 80/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 9.0908e-04 - mae: 0.0220 - r2_keras: 0.7447 - val_loss: 0.0182 - val_mae: 0.1277 - val_r2_keras: -5.1951\n",
      "Epoch 81/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 9.8686e-04 - mae: 0.0243 - r2_keras: 0.9792 - val_loss: 0.0189 - val_mae: 0.1303 - val_r2_keras: -5.4174\n",
      "Epoch 82/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - mae: 0.0272 - r2_keras: 0.9509 - val_loss: 0.0180 - val_mae: 0.1269 - val_r2_keras: -5.1296\n",
      "Epoch 83/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 8.7660e-04 - mae: 0.0227 - r2_keras: 0.9767 - val_loss: 0.0159 - val_mae: 0.1183 - val_r2_keras: -4.4174\n",
      "Epoch 84/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 8.7774e-04 - mae: 0.0234 - r2_keras: 0.6955 - val_loss: 0.0168 - val_mae: 0.1220 - val_r2_keras: -4.7007\n",
      "Epoch 85/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0255 - r2_keras: 0.6715 - val_loss: 0.0177 - val_mae: 0.1259 - val_r2_keras: -5.0204\n",
      "Epoch 86/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0014 - mae: 0.0296 - r2_keras: 0.9670 - val_loss: 0.0148 - val_mae: 0.1132 - val_r2_keras: -4.0179\n",
      "Epoch 87/100\n",
      "4/4 [==============================] - 0s 19ms/step - loss: 0.0013 - mae: 0.0280 - r2_keras: -4.7771 - val_loss: 0.0141 - val_mae: 0.1101 - val_r2_keras: -3.7889\n",
      "Epoch 88/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 8.5862e-04 - mae: 0.0229 - r2_keras: 0.8805 - val_loss: 0.0164 - val_mae: 0.1207 - val_r2_keras: -4.5949\n",
      "Epoch 89/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0244 - r2_keras: 0.9751 - val_loss: 0.0155 - val_mae: 0.1165 - val_r2_keras: -4.2563\n",
      "Epoch 90/100\n",
      "4/4 [==============================] - 0s 21ms/step - loss: 7.9072e-04 - mae: 0.0224 - r2_keras: 0.9842 - val_loss: 0.0134 - val_mae: 0.1071 - val_r2_keras: -3.5632\n",
      "Epoch 91/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0010 - mae: 0.0236 - r2_keras: 0.9764 - val_loss: 0.0133 - val_mae: 0.1065 - val_r2_keras: -3.5208\n",
      "Epoch 92/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 6.5530e-04 - mae: 0.0204 - r2_keras: 0.9754 - val_loss: 0.0144 - val_mae: 0.1116 - val_r2_keras: -3.8844\n",
      "Epoch 93/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 9.3287e-04 - mae: 0.0240 - r2_keras: 0.9547 - val_loss: 0.0175 - val_mae: 0.1253 - val_r2_keras: -4.9493\n",
      "Epoch 94/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0012 - mae: 0.0237 - r2_keras: 0.9766 - val_loss: 0.0173 - val_mae: 0.1246 - val_r2_keras: -4.8944\n",
      "Epoch 95/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 0.0014 - mae: 0.0263 - r2_keras: 0.9701 - val_loss: 0.0146 - val_mae: 0.1127 - val_r2_keras: -3.9621\n",
      "Epoch 96/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0247 - r2_keras: 0.7616 - val_loss: 0.0122 - val_mae: 0.1015 - val_r2_keras: -3.1645\n",
      "Epoch 97/100\n",
      "4/4 [==============================] - 0s 24ms/step - loss: 9.8065e-04 - mae: 0.0239 - r2_keras: -16.2309 - val_loss: 0.0107 - val_mae: 0.0937 - val_r2_keras: -2.6567\n",
      "Epoch 98/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - mae: 0.0273 - r2_keras: 0.9728 - val_loss: 0.0129 - val_mae: 0.1047 - val_r2_keras: -3.3730\n",
      "Epoch 99/100\n",
      "4/4 [==============================] - 0s 22ms/step - loss: 0.0013 - mae: 0.0305 - r2_keras: 0.9706 - val_loss: 0.0151 - val_mae: 0.1150 - val_r2_keras: -4.1347\n",
      "Epoch 100/100\n",
      "4/4 [==============================] - 0s 23ms/step - loss: 9.8269e-04 - mae: 0.0251 - r2_keras: 0.9054 - val_loss: 0.0141 - val_mae: 0.1102 - val_r2_keras: -3.7819\n"
     ]
    }
   ],
   "source": [
    "from keras.callbacks import EarlyStopping, ModelCheckpoint\n",
    "model_path = './model.h5'\n",
    "# TRAIN\n",
    "INPUT_DIMS = 4\n",
    "TIME_STEPS = 10\n",
    "lstm_units = 64\n",
    "\n",
    "#Normalized\n",
    "train,normalize = NormalizeMult(train)\n",
    "capacity_data = train[:,0].reshape(len(train),1)\n",
    "\n",
    "train_X, _ = create_dataset(train,TIME_STEPS)\n",
    "_ , train_Y = create_dataset(capacity_data,TIME_STEPS)\n",
    "print(train_X.shape,train_Y.shape)\n",
    "\n",
    "m = attention_model()\n",
    "m.summary()\n",
    "#m.compile(optimizer='adam', loss='mse')\n",
    "m.compile(loss='mse', optimizer='adam',metrics=['mae',r2_keras])\n",
    "\n",
    "# fit the network\n",
    "history =  m.fit([train_X], train_Y, epochs=100, batch_size=32, validation_split=0.2)\n",
    "#history = m.fit([train_X], train_Y, epochs=10, batch_size=64, validation_split=0.05, verbose=2,\n",
    "#          callbacks = [EarlyStopping(monitor='val_loss', min_delta=0, patience=10, verbose=0, mode='min'),\n",
    "#                       ModelCheckpoint(model_path,monitor='val_loss', save_best_only=True, mode='min', verbose=0)]\n",
    "#          )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "d65eb048-9485-43ee-8c25-5f6a809aceb8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# summarize history for R^2\n",
    "#fig_acc = plt.figure(figsize=(10, 10))\n",
    "#plt.plot(history.history['r2_keras'])\n",
    "#plt.plot(history.history['val_r2_keras'])\n",
    "#plt.title('model r^2')\n",
    "#plt.ylabel('R^2')\n",
    "#plt.xlabel('epoch')\n",
    "#plt.legend(['train', 'validation'], loc='upper left')\n",
    "#plt.show()\n",
    "\n",
    "# summarize history for MAE\n",
    "fig_acc = plt.figure(figsize=(10, 10))\n",
    "plt.plot(history.history['mae'])\n",
    "plt.plot(history.history['val_mae'])\n",
    "plt.title('model MAE')\n",
    "plt.ylabel('MAE')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'validation'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# summarize history for Loss\n",
    "fig_acc = plt.figure(figsize=(10, 10))\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'validation'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "7d770b7c-5978-421a-82d7-6402b7dc9444",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(4, 2)\n",
      "(23, 10, 4) (23, 1)\n"
     ]
    }
   ],
   "source": [
    "# TEST\n",
    "test,normalize = NormalizeMult(test)\n",
    "capacity_test = test[:,0].reshape(len(test),1)\n",
    "\n",
    "test_X, _ = create_dataset(test,TIME_STEPS)\n",
    "_ , test_Y = create_dataset(capacity_test,TIME_STEPS)\n",
    "print(test_X.shape,test_Y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "e6ae84ec-118c-41a6-aee4-d9db23869d19",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 - 0s - loss: 0.0240 - mae: 0.1287 - r2_keras: 0.4052\n"
     ]
    }
   ],
   "source": [
    "scores_test = m.evaluate([test_X], test_Y, verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "25d1dead-a68d-478b-bb61-974d11b280b8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.024031419306993484, 0.1287410706281662, 0.40522247552871704]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scores_test\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "a9d86434-70c2-4518-8318-50ab1120d5b3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.28396305],\n",
       "       [0.31514126],\n",
       "       [0.34219107],\n",
       "       [0.36849758],\n",
       "       [0.39884624],\n",
       "       [0.41875497],\n",
       "       [0.43517858],\n",
       "       [0.44988433],\n",
       "       [0.47656718],\n",
       "       [0.4846484 ],\n",
       "       [0.49683622],\n",
       "       [0.5066672 ],\n",
       "       [0.53081906],\n",
       "       [0.55087644],\n",
       "       [0.5551534 ],\n",
       "       [0.5627826 ],\n",
       "       [0.57678187],\n",
       "       [0.6068031 ],\n",
       "       [0.6187724 ],\n",
       "       [0.64436126],\n",
       "       [0.6643049 ],\n",
       "       [0.6881354 ],\n",
       "       [0.6993535 ]], dtype=float32)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results = m.predict([test_X])\n",
    "results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "923b6de8-f417-4376-9bdd-ace246808d26",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.3030303 ],\n",
       "       [0.33333333],\n",
       "       [0.36363636],\n",
       "       [0.39393939],\n",
       "       [0.42424242],\n",
       "       [0.45454545],\n",
       "       [0.48484848],\n",
       "       [0.51515152],\n",
       "       [0.54545455],\n",
       "       [0.57575758],\n",
       "       [0.60606061],\n",
       "       [0.63636364],\n",
       "       [0.66666667],\n",
       "       [0.6969697 ],\n",
       "       [0.72727273],\n",
       "       [0.75757576],\n",
       "       [0.78787879],\n",
       "       [0.81818182],\n",
       "       [0.84848485],\n",
       "       [0.87878788],\n",
       "       [0.90909091],\n",
       "       [0.93939394],\n",
       "       [0.96969697]])"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_Y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "e0fe760b-2422-4300-8498-baa18dda626c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig_acc = plt.figure(figsize=(10, 10))\n",
    "plt.plot(results[:100])\n",
    "plt.plot(test_Y[:100])\n",
    "plt.title('real vs pred')\n",
    "plt.ylabel('value')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['pred', 'real'], loc='upper left')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}