Assignment3_2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import keras\n",
    "import scipy.io\n",
    "import sklearn\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import keras.backend as K\n",
    "\n",
    "from matplotlib import pyplot as plt\n",
    "\n",
    "from keras.optimizers import Adam\n",
    "from keras.models import Model\n",
    "from keras.layers import AveragePooling1D,MaxPooling1D,UpSampling1D,AveragePooling2D,MaxPooling2D,UpSampling2D\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Input,Conv1D,Conv2D,Conv2DTranspose,Reshape\n",
    "from keras.layers import Activation,Dense,Dropout,BatchNormalization\n",
    "from keras.layers import Flatten,Lambda\n",
    "from keras.utils import np_utils\n",
    "from keras.callbacks import EarlyStopping\n",
    "\n",
    "from sklearn import preprocessing,metrics\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import classification_report,confusion_matrix\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "pathAddress=\"C:\\\\Users\\\\HP\\\\Desktop\\\\KaunsiElectivesLenaHai\\\\Neural Network and fuzzy logic BITS F312\"\n",
    "arr=scipy.io.loadmat(pathAddress+\"\\\\data_for_cnn.mat\")\n",
    "df=arr['ecg_in_window']\n",
    "label=scipy.io.loadmat(pathAddress+\"\\\\class_label.mat\")\n",
    "labels=label['label']\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "normal=preprocessing.StandardScaler()\n",
    "df=normal.fit_transform(df)\n",
    "\n",
    "X=np.expand_dims(df,axis=2)\n",
    "Y=labels\n",
    "\n",
    "x_train,x_test,y_train,y_test=train_test_split(X,Y,test_size=0.2,random_state=1)\n",
    "\n",
    "#print(x_train.shape,x_test.shape)#(800, 1000, 1) (200, 1000, 1)\n",
    "#print(\"length of x_train :\",len(x_train))# 800\n",
    "#print(\"length of y_train :\",len(y_train))# 800\n",
    "\n",
    "#print(x_train)\n",
    "\n",
    "#print(y_train.shape,y_test.shape)#(800,1) (200,1)\n",
    "#print(y_train)\n",
    "y_train=y_train.reshape(y_train.shape[0],y_train.shape[1],1)\n",
    "y_test=y_test.reshape(y_test.shape[0],y_test.shape[1],1)\n",
    "#print(\"length of x_train :\",len(x_train))# 800\n",
    "#print(\"length of y_train :\",len(y_train))# 800\n",
    "#print(y_train.shape,y_test.shape)#(800, 1, 1) (200, 1, 1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import keras.backend as K\n",
    "\n",
    "#def Conv1DTranspose(input_tensor, filters, kernel_size, strides=2, padding='same'):\n",
    "#    x = Lambda(lambda x: K.expand_dims(x, axis=2))(input_tensor)\n",
    "#    x = Conv2DTranspose(filters=filters, kernel_size=(kernel_size, 1), strides=(strides, 1), padding=padding)(x)\n",
    "#    x = Lambda(lambda x: K.squeeze(x, axis=2))(x)\n",
    "#    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model(img_shape):\n",
    "        input_layer = Input(shape=img_shape)\n",
    "        \n",
    "        # encoder\n",
    "        h = Conv1D(50,1000,activation='relu',input_shape=(1000,1))(input_layer)\n",
    "        h = MaxPooling1D(pool_size=1, padding='same')(h)\n",
    "        #h = Flatten()(h)\n",
    "        #h = Dropout(0.2)(h)\n",
    "        \n",
    "        # decoder\n",
    "        h = Dense(100,activation='sigmoid')(h)\n",
    "        h = Dropout(0.2)(h)\n",
    "        h = UpSampling1D(size=1)(h)\n",
    "        \n",
    "        kernel_size=1\n",
    "        filters=1\n",
    "        strides=1\n",
    "        padding='same'\n",
    "        \n",
    "        h = Lambda(lambda x: K.expand_dims(x, axis=2))(h)\n",
    "        h = Conv2DTranspose(filters=filters, kernel_size=(kernel_size, 1), strides=(strides, 1), padding=padding)(h)\n",
    "        output_layer = Lambda(lambda x: K.squeeze(x, axis=2))(h)\n",
    "        \n",
    "       \n",
    "        return Model(input_layer, output_layer)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 1000, 1)           0         \n",
      "_________________________________________________________________\n",
      "conv1d_1 (Conv1D)            (None, 1, 50)             50050     \n",
      "_________________________________________________________________\n",
      "max_pooling1d_1 (MaxPooling1 (None, 1, 50)             0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 1, 100)            5100      \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 1, 100)            0         \n",
      "_________________________________________________________________\n",
      "up_sampling1d_1 (UpSampling1 (None, 1, 100)            0         \n",
      "_________________________________________________________________\n",
      "lambda_1 (Lambda)            (None, 1, 1, 100)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_transpose_1 (Conv2DTr (None, 1, 1, 1)           101       \n",
      "_________________________________________________________________\n",
      "lambda_2 (Lambda)            (None, 1, 1)              0         \n",
      "=================================================================\n",
      "Total params: 55,251\n",
      "Trainable params: 55,251\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "img_rows = 800\n",
    "img_cols = 1000\n",
    "channels = 1\n",
    "#img_shape = (img_rows, img_cols, channels)\n",
    "img_shape = ( img_cols, channels)\n",
    "        \n",
    "optimizer = Adam(lr=0.001)\n",
    "        \n",
    "autoencoder_model = build_model(img_shape)\n",
    "autoencoder_model.compile(loss='mse', optimizer=optimizer,metrics=['mean_squared_error','acc'])\n",
    "autoencoder_model.summary()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_model(autoencoder_model,x_train, y_train, epochs, batch_size=20):\n",
    "    history=[]\n",
    "    early_stopping = EarlyStopping(monitor='val_loss',\n",
    "                                min_delta=0,\n",
    "                                patience=5,\n",
    "                                verbose=1, \n",
    "                                mode='auto')\n",
    "    \n",
    "    \n",
    "    history.append(autoencoder_model.fit(x_train,y_train ,\n",
    "            epochs=epochs,\n",
    "            callbacks=[early_stopping]))\n",
    "    \n",
    "    print(history[0].history.keys())\n",
    "    \n",
    "    plt.plot(history[0].history['loss'])\n",
    "    \n",
    "    plt.title('Model loss')\n",
    "    plt.ylabel('Loss')\n",
    "    plt.xlabel('Epoch')\n",
    "    plt.legend(['Train', 'Test'], loc='upper left')\n",
    "    plt.show()\n",
    "    \n",
    "    plt.figure()\n",
    "    plt.plot(history[0].history['acc'])\n",
    "    \n",
    "    plt.title('Model Accuracy')\n",
    "    plt.ylabel('Loss')\n",
    "    plt.xlabel('Epoch')\n",
    "    plt.legend(['Train', 'Test'], loc='upper left')\n",
    "    plt.show()\n",
    "    \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def eval_model(autoencoder_model,x_test):\n",
    "    preds = autoencoder_model.predict(x_test)\n",
    "    return preds\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/300\n",
      "800/800 [==============================] - 1s 642us/step - loss: 0.3763 - mean_squared_error: 0.3763 - acc: 0.5275\n",
      "Epoch 2/300\n",
      "800/800 [==============================] - 0s 133us/step - loss: 0.3551 - mean_squared_error: 0.3551 - acc: 0.5512\n",
      "Epoch 3/300\n",
      "288/800 [=========>....................] - ETA: 0s - loss: 0.3103 - mean_squared_error: 0.3103 - acc: 0.5938"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\HP\\AppData\\Roaming\\Python\\Python37\\site-packages\\keras\\callbacks\\callbacks.py:846: RuntimeWarning: Early stopping conditioned on metric `val_loss` which is not available. Available metrics are: loss,mean_squared_error,acc\n",
      "  (self.monitor, ','.join(list(logs.keys()))), RuntimeWarning\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "800/800 [==============================] - 0s 161us/step - loss: 0.3050 - mean_squared_error: 0.3050 - acc: 0.5850\n",
      "Epoch 4/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.3009 - mean_squared_error: 0.3009 - acc: 0.6112\n",
      "Epoch 5/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.2615 - mean_squared_error: 0.2615 - acc: 0.6625\n",
      "Epoch 6/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.2296 - mean_squared_error: 0.2296 - acc: 0.6812\n",
      "Epoch 7/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.1911 - mean_squared_error: 0.1911 - acc: 0.7237\n",
      "Epoch 8/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.1933 - mean_squared_error: 0.1933 - acc: 0.7412\n",
      "Epoch 9/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.1511 - mean_squared_error: 0.1511 - acc: 0.8000\n",
      "Epoch 10/300\n",
      "800/800 [==============================] - 0s 102us/step - loss: 0.1410 - mean_squared_error: 0.1410 - acc: 0.8238\n",
      "Epoch 11/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.1226 - mean_squared_error: 0.1226 - acc: 0.8525\n",
      "Epoch 12/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.1093 - mean_squared_error: 0.1093 - acc: 0.8650\n",
      "Epoch 13/300\n",
      "800/800 [==============================] - 0s 110us/step - loss: 0.1043 - mean_squared_error: 0.1043 - acc: 0.8800\n",
      "Epoch 14/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0891 - mean_squared_error: 0.0891 - acc: 0.8975\n",
      "Epoch 15/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0832 - mean_squared_error: 0.0832 - acc: 0.9150\n",
      "Epoch 16/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0811 - mean_squared_error: 0.0811 - acc: 0.9225\n",
      "Epoch 17/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0716 - mean_squared_error: 0.0716 - acc: 0.9362\n",
      "Epoch 18/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0735 - mean_squared_error: 0.0735 - acc: 0.9300\n",
      "Epoch 19/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0656 - mean_squared_error: 0.0656 - acc: 0.9475\n",
      "Epoch 20/300\n",
      "800/800 [==============================] - 0s 164us/step - loss: 0.0612 - mean_squared_error: 0.0612 - acc: 0.9538\n",
      "Epoch 21/300\n",
      "800/800 [==============================] - 0s 155us/step - loss: 0.0568 - mean_squared_error: 0.0568 - acc: 0.9638\n",
      "Epoch 22/300\n",
      "800/800 [==============================] - 0s 158us/step - loss: 0.0536 - mean_squared_error: 0.0536 - acc: 0.9663\n",
      "Epoch 23/300\n",
      "800/800 [==============================] - 0s 156us/step - loss: 0.0529 - mean_squared_error: 0.0529 - acc: 0.9600\n",
      "Epoch 24/300\n",
      "800/800 [==============================] - 0s 181us/step - loss: 0.0496 - mean_squared_error: 0.0496 - acc: 0.9712\n",
      "Epoch 25/300\n",
      "800/800 [==============================] - 0s 182us/step - loss: 0.0510 - mean_squared_error: 0.0510 - acc: 0.9638\n",
      "Epoch 26/300\n",
      "800/800 [==============================] - 0s 129us/step - loss: 0.0503 - mean_squared_error: 0.0503 - acc: 0.9650\n",
      "Epoch 27/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0435 - mean_squared_error: 0.0435 - acc: 0.9850\n",
      "Epoch 28/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0425 - mean_squared_error: 0.0425 - acc: 0.9787\n",
      "Epoch 29/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0399 - mean_squared_error: 0.0399 - acc: 0.9800\n",
      "Epoch 30/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0373 - mean_squared_error: 0.0373 - acc: 0.9850\n",
      "Epoch 31/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0378 - mean_squared_error: 0.0378 - acc: 0.9912\n",
      "Epoch 32/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0363 - mean_squared_error: 0.0363 - acc: 0.9862\n",
      "Epoch 33/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0340 - mean_squared_error: 0.0340 - acc: 0.9825\n",
      "Epoch 34/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0356 - mean_squared_error: 0.0356 - acc: 0.9912\n",
      "Epoch 35/300\n",
      "800/800 [==============================] - 0s 128us/step - loss: 0.0332 - mean_squared_error: 0.0332 - acc: 0.9887\n",
      "Epoch 36/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0329 - mean_squared_error: 0.0329 - acc: 0.9875\n",
      "Epoch 37/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0305 - mean_squared_error: 0.0305 - acc: 0.9937\n",
      "Epoch 38/300\n",
      "800/800 [==============================] - 0s 108us/step - loss: 0.0276 - mean_squared_error: 0.0276 - acc: 0.9987\n",
      "Epoch 39/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0268 - mean_squared_error: 0.0268 - acc: 0.9950\n",
      "Epoch 40/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0267 - mean_squared_error: 0.0267 - acc: 0.9937\n",
      "Epoch 41/300\n",
      "800/800 [==============================] - 0s 127us/step - loss: 0.0272 - mean_squared_error: 0.0272 - acc: 0.9975\n",
      "Epoch 42/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0263 - mean_squared_error: 0.0263 - acc: 0.9975\n",
      "Epoch 43/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0261 - mean_squared_error: 0.0261 - acc: 0.9962\n",
      "Epoch 44/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0241 - mean_squared_error: 0.0241 - acc: 0.9962\n",
      "Epoch 45/300\n",
      "800/800 [==============================] - 0s 141us/step - loss: 0.0240 - mean_squared_error: 0.0240 - acc: 0.9987\n",
      "Epoch 46/300\n",
      "800/800 [==============================] - 0s 111us/step - loss: 0.0243 - mean_squared_error: 0.0243 - acc: 0.9937\n",
      "Epoch 47/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0200 - mean_squared_error: 0.0200 - acc: 0.9987\n",
      "Epoch 48/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0210 - mean_squared_error: 0.0210 - acc: 0.9975\n",
      "Epoch 49/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0188 - mean_squared_error: 0.0188 - acc: 0.9987\n",
      "Epoch 50/300\n",
      "800/800 [==============================] - 0s 112us/step - loss: 0.0220 - mean_squared_error: 0.0220 - acc: 0.9975\n",
      "Epoch 51/300\n",
      "800/800 [==============================] - 0s 148us/step - loss: 0.0189 - mean_squared_error: 0.0189 - acc: 1.0000\n",
      "Epoch 52/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0177 - mean_squared_error: 0.0177 - acc: 1.0000\n",
      "Epoch 53/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0172 - mean_squared_error: 0.0172 - acc: 1.0000\n",
      "Epoch 54/300\n",
      "800/800 [==============================] - 0s 107us/step - loss: 0.0189 - mean_squared_error: 0.0189 - acc: 1.0000\n",
      "Epoch 55/300\n",
      "800/800 [==============================] - 0s 111us/step - loss: 0.0181 - mean_squared_error: 0.0181 - acc: 0.9987\n",
      "Epoch 56/300\n",
      "800/800 [==============================] - 0s 112us/step - loss: 0.0177 - mean_squared_error: 0.0177 - acc: 0.9987\n",
      "Epoch 57/300\n",
      "800/800 [==============================] - 0s 132us/step - loss: 0.0172 - mean_squared_error: 0.0172 - acc: 1.0000\n",
      "Epoch 58/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0185 - mean_squared_error: 0.0185 - acc: 0.9975\n",
      "Epoch 59/300\n",
      "800/800 [==============================] - 0s 112us/step - loss: 0.0156 - mean_squared_error: 0.0156 - acc: 0.9987\n",
      "Epoch 60/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0138 - mean_squared_error: 0.0138 - acc: 1.0000\n",
      "Epoch 61/300\n",
      "800/800 [==============================] - 0s 112us/step - loss: 0.0162 - mean_squared_error: 0.0162 - acc: 1.0000\n",
      "Epoch 62/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0161 - mean_squared_error: 0.0161 - acc: 1.0000\n",
      "Epoch 63/300\n",
      "800/800 [==============================] - 0s 105us/step - loss: 0.0150 - mean_squared_error: 0.0150 - acc: 1.0000\n",
      "Epoch 64/300\n",
      "800/800 [==============================] - 0s 108us/step - loss: 0.0138 - mean_squared_error: 0.0138 - acc: 1.0000\n",
      "Epoch 65/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0156 - mean_squared_error: 0.0156 - acc: 1.0000\n",
      "Epoch 66/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0159 - mean_squared_error: 0.0159 - acc: 0.9987\n",
      "Epoch 67/300\n",
      "800/800 [==============================] - 0s 109us/step - loss: 0.0148 - mean_squared_error: 0.0148 - acc: 1.0000\n",
      "Epoch 68/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0134 - mean_squared_error: 0.0134 - acc: 1.0000\n",
      "Epoch 69/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0133 - mean_squared_error: 0.0133 - acc: 0.9987\n",
      "Epoch 70/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0140 - mean_squared_error: 0.0140 - acc: 0.9987\n",
      "Epoch 71/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0139 - mean_squared_error: 0.0139 - acc: 1.0000\n",
      "Epoch 72/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0123 - mean_squared_error: 0.0123 - acc: 1.0000\n",
      "Epoch 73/300\n",
      "800/800 [==============================] - 0s 128us/step - loss: 0.0123 - mean_squared_error: 0.0123 - acc: 1.0000\n",
      "Epoch 74/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0125 - mean_squared_error: 0.0125 - acc: 1.0000\n",
      "Epoch 75/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0123 - mean_squared_error: 0.0123 - acc: 1.0000\n",
      "Epoch 76/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0156 - mean_squared_error: 0.0156 - acc: 0.9987\n",
      "Epoch 77/300\n",
      "800/800 [==============================] - 0s 154us/step - loss: 0.0143 - mean_squared_error: 0.0143 - acc: 1.0000\n",
      "Epoch 78/300\n",
      "800/800 [==============================] - 0s 146us/step - loss: 0.0119 - mean_squared_error: 0.0119 - acc: 1.0000\n",
      "Epoch 79/300\n",
      "800/800 [==============================] - 0s 155us/step - loss: 0.0117 - mean_squared_error: 0.0117 - acc: 1.0000\n",
      "Epoch 80/300\n",
      "800/800 [==============================] - 0s 158us/step - loss: 0.0118 - mean_squared_error: 0.0118 - acc: 1.0000\n",
      "Epoch 81/300\n",
      "800/800 [==============================] - 0s 164us/step - loss: 0.0115 - mean_squared_error: 0.0115 - acc: 1.0000\n",
      "Epoch 82/300\n",
      "800/800 [==============================] - 0s 155us/step - loss: 0.0114 - mean_squared_error: 0.0114 - acc: 1.0000\n",
      "Epoch 83/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0096 - mean_squared_error: 0.0096 - acc: 1.0000\n",
      "Epoch 84/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0095 - mean_squared_error: 0.0095 - acc: 1.0000\n",
      "Epoch 85/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0104 - mean_squared_error: 0.0104 - acc: 1.0000\n",
      "Epoch 86/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0100 - mean_squared_error: 0.0100 - acc: 1.0000\n",
      "Epoch 87/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0102 - mean_squared_error: 0.0102 - acc: 1.0000\n",
      "Epoch 88/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0096 - mean_squared_error: 0.0096 - acc: 1.0000\n",
      "Epoch 89/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0108 - mean_squared_error: 0.0108 - acc: 1.0000\n",
      "Epoch 90/300\n",
      "800/800 [==============================] - 0s 110us/step - loss: 0.0092 - mean_squared_error: 0.0092 - acc: 1.0000\n",
      "Epoch 91/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0092 - mean_squared_error: 0.0092 - acc: 1.0000\n",
      "Epoch 92/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0090 - mean_squared_error: 0.0090 - acc: 1.0000\n",
      "Epoch 93/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0098 - mean_squared_error: 0.0098 - acc: 1.0000\n",
      "Epoch 94/300\n",
      "800/800 [==============================] - 0s 132us/step - loss: 0.0099 - mean_squared_error: 0.0099 - acc: 1.0000\n",
      "Epoch 95/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0100 - mean_squared_error: 0.0100 - acc: 1.0000\n",
      "Epoch 96/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0104 - mean_squared_error: 0.0104 - acc: 1.0000\n",
      "Epoch 97/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0100 - mean_squared_error: 0.0100 - acc: 1.0000\n",
      "Epoch 98/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0091 - mean_squared_error: 0.0091 - acc: 1.0000\n",
      "Epoch 99/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0108 - mean_squared_error: 0.0108 - acc: 1.0000\n",
      "Epoch 100/300\n",
      "800/800 [==============================] - 0s 152us/step - loss: 0.0091 - mean_squared_error: 0.0091 - acc: 1.0000\n",
      "Epoch 101/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0092 - mean_squared_error: 0.0092 - acc: 1.0000\n",
      "Epoch 102/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0090 - mean_squared_error: 0.0090 - acc: 1.0000\n",
      "Epoch 103/300\n",
      "800/800 [==============================] - 0s 112us/step - loss: 0.0082 - mean_squared_error: 0.0082 - acc: 1.0000\n",
      "Epoch 104/300\n",
      "800/800 [==============================] - 0s 109us/step - loss: 0.0085 - mean_squared_error: 0.0085 - acc: 1.0000\n",
      "Epoch 105/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0078 - mean_squared_error: 0.0078 - acc: 1.0000\n",
      "Epoch 106/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0079 - mean_squared_error: 0.0079 - acc: 1.0000\n",
      "Epoch 107/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0079 - mean_squared_error: 0.0079 - acc: 1.0000\n",
      "Epoch 108/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0084 - mean_squared_error: 0.0084 - acc: 1.0000\n",
      "Epoch 109/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0086 - mean_squared_error: 0.0086 - acc: 1.0000\n",
      "Epoch 110/300\n",
      "800/800 [==============================] - 0s 122us/step - loss: 0.0076 - mean_squared_error: 0.0076 - acc: 1.0000\n",
      "Epoch 111/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0075 - mean_squared_error: 0.0075 - acc: 1.0000\n",
      "Epoch 112/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0091 - mean_squared_error: 0.0091 - acc: 1.0000\n",
      "Epoch 113/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0085 - mean_squared_error: 0.0085 - acc: 1.0000\n",
      "Epoch 114/300\n",
      "800/800 [==============================] - 0s 109us/step - loss: 0.0077 - mean_squared_error: 0.0077 - acc: 1.0000\n",
      "Epoch 115/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0079 - mean_squared_error: 0.0079 - acc: 1.0000\n",
      "Epoch 116/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0080 - mean_squared_error: 0.0080 - acc: 1.0000\n",
      "Epoch 117/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0076 - mean_squared_error: 0.0076 - acc: 1.0000\n",
      "Epoch 118/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0081 - mean_squared_error: 0.0081 - acc: 1.0000\n",
      "Epoch 119/300\n",
      "800/800 [==============================] - 0s 132us/step - loss: 0.0073 - mean_squared_error: 0.0073 - acc: 1.0000\n",
      "Epoch 120/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0086 - mean_squared_error: 0.0086 - acc: 1.0000\n",
      "Epoch 121/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0070 - mean_squared_error: 0.0070 - acc: 1.0000\n",
      "Epoch 122/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0086 - mean_squared_error: 0.0086 - acc: 1.0000\n",
      "Epoch 123/300\n",
      "800/800 [==============================] - 0s 112us/step - loss: 0.0088 - mean_squared_error: 0.0088 - acc: 1.0000\n",
      "Epoch 124/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0069 - mean_squared_error: 0.0069 - acc: 1.0000\n",
      "Epoch 125/300\n",
      "800/800 [==============================] - 0s 105us/step - loss: 0.0074 - mean_squared_error: 0.0074 - acc: 1.0000\n",
      "Epoch 126/300\n",
      "800/800 [==============================] - 0s 112us/step - loss: 0.0074 - mean_squared_error: 0.0074 - acc: 1.0000\n",
      "Epoch 127/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0074 - mean_squared_error: 0.0074 - acc: 1.0000\n",
      "Epoch 128/300\n",
      "800/800 [==============================] - 0s 110us/step - loss: 0.0070 - mean_squared_error: 0.0070 - acc: 1.0000\n",
      "Epoch 129/300\n",
      "800/800 [==============================] - 0s 110us/step - loss: 0.0074 - mean_squared_error: 0.0074 - acc: 1.0000\n",
      "Epoch 130/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0069 - mean_squared_error: 0.0069 - acc: 1.0000\n",
      "Epoch 131/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0066 - mean_squared_error: 0.0066 - acc: 1.0000\n",
      "Epoch 132/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0075 - mean_squared_error: 0.0075 - acc: 1.0000\n",
      "Epoch 133/300\n",
      "800/800 [==============================] - 0s 129us/step - loss: 0.0074 - mean_squared_error: 0.0074 - acc: 0.9987\n",
      "Epoch 134/300\n",
      "800/800 [==============================] - 0s 158us/step - loss: 0.0073 - mean_squared_error: 0.0073 - acc: 1.0000\n",
      "Epoch 135/300\n",
      "800/800 [==============================] - 0s 134us/step - loss: 0.0075 - mean_squared_error: 0.0075 - acc: 1.0000\n",
      "Epoch 136/300\n",
      "800/800 [==============================] - 0s 166us/step - loss: 0.0066 - mean_squared_error: 0.0066 - acc: 1.0000\n",
      "Epoch 137/300\n",
      "800/800 [==============================] - 0s 147us/step - loss: 0.0065 - mean_squared_error: 0.0065 - acc: 1.0000\n",
      "Epoch 138/300\n",
      "800/800 [==============================] - 0s 161us/step - loss: 0.0074 - mean_squared_error: 0.0074 - acc: 1.0000\n",
      "Epoch 139/300\n",
      "800/800 [==============================] - 0s 145us/step - loss: 0.0064 - mean_squared_error: 0.0064 - acc: 1.0000\n",
      "Epoch 140/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0068 - mean_squared_error: 0.0068 - acc: 1.0000\n",
      "Epoch 141/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0059 - mean_squared_error: 0.0059 - acc: 1.0000\n",
      "Epoch 142/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0060 - mean_squared_error: 0.0060 - acc: 1.0000\n",
      "Epoch 143/300\n",
      "800/800 [==============================] - 0s 111us/step - loss: 0.0063 - mean_squared_error: 0.0063 - acc: 1.0000\n",
      "Epoch 144/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0068 - mean_squared_error: 0.0068 - acc: 1.0000\n",
      "Epoch 145/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0065 - mean_squared_error: 0.0065 - acc: 1.0000\n",
      "Epoch 146/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0071 - mean_squared_error: 0.0071 - acc: 1.0000\n",
      "Epoch 147/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0065 - mean_squared_error: 0.0065 - acc: 1.0000\n",
      "Epoch 148/300\n",
      "800/800 [==============================] - 0s 112us/step - loss: 0.0067 - mean_squared_error: 0.0067 - acc: 1.0000\n",
      "Epoch 149/300\n",
      "800/800 [==============================] - 0s 156us/step - loss: 0.0066 - mean_squared_error: 0.0066 - acc: 1.0000\n",
      "Epoch 150/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0057 - mean_squared_error: 0.0057 - acc: 1.0000\n",
      "Epoch 151/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0057 - mean_squared_error: 0.0057 - acc: 1.0000\n",
      "Epoch 152/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0065 - mean_squared_error: 0.0065 - acc: 1.0000\n",
      "Epoch 153/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0061 - mean_squared_error: 0.0061 - acc: 1.0000\n",
      "Epoch 154/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0051 - mean_squared_error: 0.0051 - acc: 1.0000\n",
      "Epoch 155/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0059 - mean_squared_error: 0.0059 - acc: 1.0000\n",
      "Epoch 156/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0058 - mean_squared_error: 0.0058 - acc: 1.0000\n",
      "Epoch 157/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0062 - mean_squared_error: 0.0062 - acc: 1.0000\n",
      "Epoch 158/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0059 - mean_squared_error: 0.0059 - acc: 1.0000\n",
      "Epoch 159/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0055 - mean_squared_error: 0.0055 - acc: 1.0000\n",
      "Epoch 160/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0054 - mean_squared_error: 0.0054 - acc: 1.0000\n",
      "Epoch 161/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0058 - mean_squared_error: 0.0058 - acc: 1.0000\n",
      "Epoch 162/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0058 - mean_squared_error: 0.0058 - acc: 1.0000\n",
      "Epoch 163/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0053 - mean_squared_error: 0.0053 - acc: 1.0000\n",
      "Epoch 164/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0054 - mean_squared_error: 0.0054 - acc: 1.0000\n",
      "Epoch 165/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0066 - mean_squared_error: 0.0066 - acc: 1.0000\n",
      "Epoch 166/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0062 - mean_squared_error: 0.0062 - acc: 1.0000\n",
      "Epoch 167/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0064 - mean_squared_error: 0.0064 - acc: 1.0000\n",
      "Epoch 168/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0058 - mean_squared_error: 0.0058 - acc: 1.0000\n",
      "Epoch 169/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0063 - mean_squared_error: 0.0063 - acc: 1.0000\n",
      "Epoch 170/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0054 - mean_squared_error: 0.0054 - acc: 1.0000\n",
      "Epoch 171/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0052 - mean_squared_error: 0.0052 - acc: 1.0000\n",
      "Epoch 172/300\n",
      "800/800 [==============================] - 0s 131us/step - loss: 0.0053 - mean_squared_error: 0.0053 - acc: 1.0000\n",
      "Epoch 173/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0057 - mean_squared_error: 0.0057 - acc: 1.0000\n",
      "Epoch 174/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0062 - mean_squared_error: 0.0062 - acc: 1.0000\n",
      "Epoch 175/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0062 - mean_squared_error: 0.0062 - acc: 1.0000\n",
      "Epoch 176/300\n",
      "800/800 [==============================] - 0s 134us/step - loss: 0.0066 - mean_squared_error: 0.0066 - acc: 1.0000\n",
      "Epoch 177/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0054 - mean_squared_error: 0.0054 - acc: 1.0000\n",
      "Epoch 178/300\n",
      "800/800 [==============================] - 0s 127us/step - loss: 0.0055 - mean_squared_error: 0.0055 - acc: 1.0000\n",
      "Epoch 179/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0049 - mean_squared_error: 0.0049 - acc: 1.0000\n",
      "Epoch 180/300\n",
      "800/800 [==============================] - 0s 127us/step - loss: 0.0050 - mean_squared_error: 0.0050 - acc: 1.0000\n",
      "Epoch 181/300\n",
      "800/800 [==============================] - 0s 126us/step - loss: 0.0057 - mean_squared_error: 0.0057 - acc: 1.0000\n",
      "Epoch 182/300\n",
      "800/800 [==============================] - 0s 130us/step - loss: 0.0051 - mean_squared_error: 0.0051 - acc: 1.0000\n",
      "Epoch 183/300\n",
      "800/800 [==============================] - 0s 129us/step - loss: 0.0048 - mean_squared_error: 0.0048 - acc: 1.0000\n",
      "Epoch 184/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0052 - mean_squared_error: 0.0052 - acc: 1.0000\n",
      "Epoch 185/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0049 - mean_squared_error: 0.0049 - acc: 1.0000\n",
      "Epoch 186/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0054 - mean_squared_error: 0.0054 - acc: 1.0000\n",
      "Epoch 187/300\n",
      "800/800 [==============================] - 0s 129us/step - loss: 0.0057 - mean_squared_error: 0.0057 - acc: 1.0000\n",
      "Epoch 188/300\n",
      "800/800 [==============================] - 0s 137us/step - loss: 0.0055 - mean_squared_error: 0.0055 - acc: 1.0000\n",
      "Epoch 189/300\n",
      "800/800 [==============================] - 0s 155us/step - loss: 0.0050 - mean_squared_error: 0.0050 - acc: 1.0000\n",
      "Epoch 190/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0049 - mean_squared_error: 0.0049 - acc: 1.0000\n",
      "Epoch 191/300\n",
      "800/800 [==============================] - 0s 156us/step - loss: 0.0051 - mean_squared_error: 0.0051 - acc: 1.0000\n",
      "Epoch 192/300\n",
      "800/800 [==============================] - 0s 131us/step - loss: 0.0055 - mean_squared_error: 0.0055 - acc: 1.0000\n",
      "Epoch 193/300\n",
      "800/800 [==============================] - 0s 177us/step - loss: 0.0048 - mean_squared_error: 0.0048 - acc: 1.0000\n",
      "Epoch 194/300\n",
      "800/800 [==============================] - 0s 187us/step - loss: 0.0049 - mean_squared_error: 0.0049 - acc: 1.0000\n",
      "Epoch 195/300\n",
      "800/800 [==============================] - 0s 151us/step - loss: 0.0044 - mean_squared_error: 0.0044 - acc: 1.0000\n",
      "Epoch 196/300\n",
      "800/800 [==============================] - 0s 163us/step - loss: 0.0046 - mean_squared_error: 0.0046 - acc: 1.0000\n",
      "Epoch 197/300\n",
      "800/800 [==============================] - 0s 131us/step - loss: 0.0048 - mean_squared_error: 0.0048 - acc: 1.0000\n",
      "Epoch 198/300\n",
      "800/800 [==============================] - 0s 132us/step - loss: 0.0045 - mean_squared_error: 0.0045 - acc: 1.0000\n",
      "Epoch 199/300\n",
      "800/800 [==============================] - 0s 128us/step - loss: 0.0042 - mean_squared_error: 0.0042 - acc: 1.0000\n",
      "Epoch 200/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0043 - mean_squared_error: 0.0043 - acc: 1.0000\n",
      "Epoch 201/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0048 - mean_squared_error: 0.0048 - acc: 1.0000\n",
      "Epoch 202/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0045 - mean_squared_error: 0.0045 - acc: 1.0000\n",
      "Epoch 203/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0043 - mean_squared_error: 0.0043 - acc: 1.0000\n",
      "Epoch 204/300\n",
      "800/800 [==============================] - 0s 122us/step - loss: 0.0056 - mean_squared_error: 0.0056 - acc: 1.0000\n",
      "Epoch 205/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0051 - mean_squared_error: 0.0051 - acc: 1.0000\n",
      "Epoch 206/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0053 - mean_squared_error: 0.0053 - acc: 1.0000\n",
      "Epoch 207/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0045 - mean_squared_error: 0.0045 - acc: 1.0000\n",
      "Epoch 208/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0046 - mean_squared_error: 0.0046 - acc: 1.0000\n",
      "Epoch 209/300\n",
      "800/800 [==============================] - 0s 128us/step - loss: 0.0044 - mean_squared_error: 0.0044 - acc: 1.0000\n",
      "Epoch 210/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0045 - mean_squared_error: 0.0045 - acc: 1.0000\n",
      "Epoch 211/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0045 - mean_squared_error: 0.0045 - acc: 1.0000\n",
      "Epoch 212/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0044 - mean_squared_error: 0.0044 - acc: 1.0000\n",
      "Epoch 213/300\n",
      "800/800 [==============================] - 0s 122us/step - loss: 0.0044 - mean_squared_error: 0.0044 - acc: 1.0000\n",
      "Epoch 214/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0043 - mean_squared_error: 0.0043 - acc: 1.0000\n",
      "Epoch 215/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0041 - mean_squared_error: 0.0041 - acc: 1.0000\n",
      "Epoch 216/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0043 - mean_squared_error: 0.0043 - acc: 1.0000\n",
      "Epoch 217/300\n",
      "800/800 [==============================] - 0s 122us/step - loss: 0.0047 - mean_squared_error: 0.0047 - acc: 1.0000\n",
      "Epoch 218/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0079 - mean_squared_error: 0.0079 - acc: 0.9962\n",
      "Epoch 219/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0135 - mean_squared_error: 0.0135 - acc: 0.9937\n",
      "Epoch 220/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0066 - mean_squared_error: 0.0066 - acc: 1.0000\n",
      "Epoch 221/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0049 - mean_squared_error: 0.0049 - acc: 1.0000\n",
      "Epoch 222/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0051 - mean_squared_error: 0.0051 - acc: 1.0000\n",
      "Epoch 223/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0041 - mean_squared_error: 0.0041 - acc: 1.0000\n",
      "Epoch 224/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0043 - mean_squared_error: 0.0043 - acc: 1.0000\n",
      "Epoch 225/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0036 - mean_squared_error: 0.0036 - acc: 1.0000\n",
      "Epoch 226/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0037 - mean_squared_error: 0.0037 - acc: 1.0000\n",
      "Epoch 227/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0039 - mean_squared_error: 0.0039 - acc: 1.0000\n",
      "Epoch 228/300\n",
      "800/800 [==============================] - 0s 111us/step - loss: 0.0035 - mean_squared_error: 0.0035 - acc: 1.0000\n",
      "Epoch 229/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0036 - mean_squared_error: 0.0036 - acc: 1.0000\n",
      "Epoch 230/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0037 - mean_squared_error: 0.0037 - acc: 1.0000\n",
      "Epoch 231/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0036 - mean_squared_error: 0.0036 - acc: 1.0000\n",
      "Epoch 232/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0037 - mean_squared_error: 0.0037 - acc: 1.0000\n",
      "Epoch 233/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0037 - mean_squared_error: 0.0037 - acc: 1.0000\n",
      "Epoch 234/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0040 - mean_squared_error: 0.0040 - acc: 1.0000\n",
      "Epoch 235/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0044 - mean_squared_error: 0.0044 - acc: 1.0000\n",
      "Epoch 236/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0040 - mean_squared_error: 0.0040 - acc: 1.0000\n",
      "Epoch 237/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0039 - mean_squared_error: 0.0039 - acc: 1.0000\n",
      "Epoch 238/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0040 - mean_squared_error: 0.0040 - acc: 1.0000\n",
      "Epoch 239/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0039 - mean_squared_error: 0.0039 - acc: 1.0000\n",
      "Epoch 240/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0041 - mean_squared_error: 0.0041 - acc: 1.0000\n",
      "Epoch 241/300\n",
      "800/800 [==============================] - 0s 111us/step - loss: 0.0035 - mean_squared_error: 0.0035 - acc: 1.0000\n",
      "Epoch 242/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0041 - mean_squared_error: 0.0041 - acc: 1.0000\n",
      "Epoch 243/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0039 - mean_squared_error: 0.0039 - acc: 1.0000\n",
      "Epoch 244/300\n",
      "800/800 [==============================] - 0s 122us/step - loss: 0.0040 - mean_squared_error: 0.0040 - acc: 1.0000\n",
      "Epoch 245/300\n",
      "800/800 [==============================] - 0s 144us/step - loss: 0.0036 - mean_squared_error: 0.0036 - acc: 1.0000\n",
      "Epoch 246/300\n",
      "800/800 [==============================] - 0s 165us/step - loss: 0.0034 - mean_squared_error: 0.0034 - acc: 1.0000\n",
      "Epoch 247/300\n",
      "800/800 [==============================] - 0s 132us/step - loss: 0.0035 - mean_squared_error: 0.0035 - acc: 1.0000\n",
      "Epoch 248/300\n",
      "800/800 [==============================] - 0s 147us/step - loss: 0.0040 - mean_squared_error: 0.0040 - acc: 1.0000\n",
      "Epoch 249/300\n",
      "800/800 [==============================] - 0s 155us/step - loss: 0.0042 - mean_squared_error: 0.0042 - acc: 1.0000\n",
      "Epoch 250/300\n",
      "800/800 [==============================] - 0s 127us/step - loss: 0.0041 - mean_squared_error: 0.0041 - acc: 1.0000\n",
      "Epoch 251/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0033 - mean_squared_error: 0.0033 - acc: 1.0000\n",
      "Epoch 252/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0031 - mean_squared_error: 0.0031 - acc: 1.0000\n",
      "Epoch 253/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0057 - mean_squared_error: 0.0057 - acc: 0.9987\n",
      "Epoch 254/300\n",
      "800/800 [==============================] - 0s 113us/step - loss: 0.0085 - mean_squared_error: 0.0085 - acc: 0.9962\n",
      "Epoch 255/300\n",
      "800/800 [==============================] - 0s 112us/step - loss: 0.0117 - mean_squared_error: 0.0117 - acc: 0.9912\n",
      "Epoch 256/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0105 - mean_squared_error: 0.0105 - acc: 0.9962\n",
      "Epoch 257/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0122 - mean_squared_error: 0.0122 - acc: 0.9900\n",
      "Epoch 258/300\n",
      "800/800 [==============================] - 0s 122us/step - loss: 0.0174 - mean_squared_error: 0.0174 - acc: 0.9887\n",
      "Epoch 259/300\n",
      "800/800 [==============================] - 0s 127us/step - loss: 0.0089 - mean_squared_error: 0.0089 - acc: 0.9987\n",
      "Epoch 260/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0048 - mean_squared_error: 0.0048 - acc: 1.0000\n",
      "Epoch 261/300\n",
      "800/800 [==============================] - 0s 128us/step - loss: 0.0050 - mean_squared_error: 0.0050 - acc: 1.0000\n",
      "Epoch 262/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0062 - mean_squared_error: 0.0062 - acc: 1.0000\n",
      "Epoch 263/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0063 - mean_squared_error: 0.0063 - acc: 0.9987\n",
      "Epoch 264/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0047 - mean_squared_error: 0.0047 - acc: 1.0000\n",
      "Epoch 265/300\n",
      "800/800 [==============================] - 0s 127us/step - loss: 0.0046 - mean_squared_error: 0.0046 - acc: 1.0000\n",
      "Epoch 266/300\n",
      "800/800 [==============================] - 0s 126us/step - loss: 0.0036 - mean_squared_error: 0.0036 - acc: 1.0000\n",
      "Epoch 267/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0035 - mean_squared_error: 0.0035 - acc: 1.0000\n",
      "Epoch 268/300\n",
      "800/800 [==============================] - 0s 122us/step - loss: 0.0034 - mean_squared_error: 0.0034 - acc: 1.0000\n",
      "Epoch 269/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0035 - mean_squared_error: 0.0035 - acc: 1.0000\n",
      "Epoch 270/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0031 - mean_squared_error: 0.0031 - acc: 1.0000\n",
      "Epoch 271/300\n",
      "800/800 [==============================] - 0s 122us/step - loss: 0.0033 - mean_squared_error: 0.0033 - acc: 1.0000\n",
      "Epoch 272/300\n",
      "800/800 [==============================] - 0s 122us/step - loss: 0.0032 - mean_squared_error: 0.0032 - acc: 1.0000\n",
      "Epoch 273/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0036 - mean_squared_error: 0.0036 - acc: 1.0000\n",
      "Epoch 274/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0030 - mean_squared_error: 0.0030 - acc: 1.0000\n",
      "Epoch 275/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0033 - mean_squared_error: 0.0033 - acc: 1.0000\n",
      "Epoch 276/300\n",
      "800/800 [==============================] - 0s 115us/step - loss: 0.0034 - mean_squared_error: 0.0034 - acc: 1.0000\n",
      "Epoch 277/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0036 - mean_squared_error: 0.0036 - acc: 1.0000\n",
      "Epoch 278/300\n",
      "800/800 [==============================] - 0s 117us/step - loss: 0.0030 - mean_squared_error: 0.0030 - acc: 1.0000\n",
      "Epoch 279/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0032 - mean_squared_error: 0.0032 - acc: 1.0000\n",
      "Epoch 280/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0033 - mean_squared_error: 0.0033 - acc: 1.0000\n",
      "Epoch 281/300\n",
      "800/800 [==============================] - 0s 121us/step - loss: 0.0031 - mean_squared_error: 0.0031 - acc: 1.0000\n",
      "Epoch 282/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0031 - mean_squared_error: 0.0031 - acc: 1.0000\n",
      "Epoch 283/300\n",
      "800/800 [==============================] - 0s 124us/step - loss: 0.0029 - mean_squared_error: 0.0029 - acc: 1.0000\n",
      "Epoch 284/300\n",
      "800/800 [==============================] - 0s 123us/step - loss: 0.0030 - mean_squared_error: 0.0030 - acc: 1.0000\n",
      "Epoch 285/300\n",
      "800/800 [==============================] - 0s 120us/step - loss: 0.0032 - mean_squared_error: 0.0032 - acc: 1.0000\n",
      "Epoch 286/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0031 - mean_squared_error: 0.0031 - acc: 1.0000\n",
      "Epoch 287/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0034 - mean_squared_error: 0.0034 - acc: 1.0000\n",
      "Epoch 288/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0032 - mean_squared_error: 0.0032 - acc: 1.0000\n",
      "Epoch 289/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0029 - mean_squared_error: 0.0029 - acc: 1.0000\n",
      "Epoch 290/300\n",
      "800/800 [==============================] - 0s 114us/step - loss: 0.0030 - mean_squared_error: 0.0030 - acc: 1.0000\n",
      "Epoch 291/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0035 - mean_squared_error: 0.0035 - acc: 1.0000\n",
      "Epoch 292/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0031 - mean_squared_error: 0.0031 - acc: 1.0000\n",
      "Epoch 293/300\n",
      "800/800 [==============================] - 0s 122us/step - loss: 0.0033 - mean_squared_error: 0.0033 - acc: 1.0000\n",
      "Epoch 294/300\n",
      "800/800 [==============================] - 0s 118us/step - loss: 0.0032 - mean_squared_error: 0.0032 - acc: 1.0000\n",
      "Epoch 295/300\n",
      "800/800 [==============================] - 0s 135us/step - loss: 0.0027 - mean_squared_error: 0.0027 - acc: 1.0000\n",
      "Epoch 296/300\n",
      "800/800 [==============================] - 0s 116us/step - loss: 0.0035 - mean_squared_error: 0.0035 - acc: 1.0000\n",
      "Epoch 297/300\n",
      "800/800 [==============================] - 0s 111us/step - loss: 0.0030 - mean_squared_error: 0.0030 - acc: 1.0000\n",
      "Epoch 298/300\n",
      "800/800 [==============================] - 0s 110us/step - loss: 0.0030 - mean_squared_error: 0.0030 - acc: 1.0000\n",
      "Epoch 299/300\n",
      "800/800 [==============================] - 0s 119us/step - loss: 0.0031 - mean_squared_error: 0.0031 - acc: 1.0000\n",
      "Epoch 300/300\n",
      "800/800 [==============================] - 0s 125us/step - loss: 0.0030 - mean_squared_error: 0.0030 - acc: 1.0000\n",
      "dict_keys(['loss', 'mean_squared_error', 'acc'])\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"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3de3xdZZ3v8c8vO9c2SS9J2kLvhYIt6ABWFGGAEUWKCt6FI4oeBjweUdRRB0ZFYPTozHg7KKMHFbyhiNdhOCg4CCoeBQoWhNZCKS0NvSRNb7nv7OR3/lgrYbO7kyYlKyvZz/f9eu1X115r7eS3stP9zfM8az3L3B0REQlXWdoFiIhIuhQEIiKBUxCIiAROQSAiEjgFgYhI4BQEIiKBUxBIEMxsiZm5mZWPYt93mdm9E1GXyGSgIJBJx8w2m1nWzBoL1q+NP8yXpFPZc2qZbmYdZnZ72rWIPF8KApmsngLOH3xiZi8EatIr5wBvBnqBM83ssIn8xqNp1YiMhYJAJqvvAe/Me34h8N38Hcxshpl918xazWyLmX3CzMribRkz+7yZ7TKzTcBrirz2W2a23cyeMbNPm1lmDPVdCHwdeAR4e8HXXmhmP4vrajOzr+Ztu9jM1ptZu5mtM7MT4vVuZkfm7fdtM/t0vHy6mTWb2T+a2Q7gRjObZWa3xd9jT7y8IO/1s83sRjPbFm//Rbz+UTN7Xd5+FfHP6LgxHLuUGAWBTFZ/AurNbEX8Af024PsF+3wFmAEsA04jCo53x9suBl4LHA+sIvoLPt93gBxwZLzPmcDfj6YwM1sEnA7cFD/embctA9wGbAGWAPOBm+NtbwGuivevB84B2kbzPYF5wGxgMXAJ0f/dG+Pni4Bu4Kt5+38PmAYcA8wBvhSv/y5wQd5+ZwPb3X3tKOuQUuTueugxqR7AZuCVwCeAzwJnAb8GygEn+oDNEHXNrMx73XuAe+Ll3wD/I2/bmfFry4G58Wtr8rafD9wdL78LuHeE+j4BrI2XDwf6gePj5ycBrUB5kdfdAVw2zNd04Mi8598GPh0vnw5kgeoRajoO2BMvHwYMALOK7Hc40A7Ux89/Anws7fdcj3Qf6muUyex7wO+ApRR0CwGNQCXRX96DthD9BQ7RB97Wgm2DFgMVwHYzG1xXVrD/SN4JfAPA3beZ2W+Juor+DCwEtrh7rsjrFgJPjvJ7FGp1957BJ2Y2jeiv/LOAWfHqurhFshDY7e57Cr9IXO8fgDeZ2c+B1cBlh1iTlAh1Dcmk5e5biAaNzwZ+VrB5F9BH9KE+aBHwTLy8negDMX/boK1ELYJGd58ZP+rd/ZiD1WRmLweWA1eY2Y64z/6lwPnxIO5WYNEwA7pbgSOG+dJdRF05g+YVbC+cJvgfgKOBl7p7PXDqYInx95ltZjOH+V7fIeoeegvwR3d/Zpj9JBAKApnsLgJe4e6d+SvdvR+4BfiMmdWZ2WLgwzw7jnAL8AEzW2Bms4DL8167HbgT+IKZ1ZtZmZkdYWanjaKeC4m6qVYSdcccBxxL9CG+GrifKIQ+F59iWm1mJ8ev/SbwETN7sUWOjOsGWAv8t3iQ+yyiMY+R1BGNC+w1s9nApwqO75fAv8eDyhVmdmrea38BnEDUEihsaUmAFAQyqbn7k+6+ZpjN7wc6gU3AvcAPgBvibd8g6pN/GHiIA1sU7yTqWloH7CHqKx/xNFAzqwbeCnzF3XfkPZ4i6sa6MA6o1xENQj8NNBMNdOPuPwY+E9fZTvSBPDv+8pfFr9tLdBbSL0aqBfgy0em0u4gG1n9VsP0dRC2mvwItwAcHN7h7N/BToi63wp+LBMjcdWMakdCY2ZXAUe5+wUF3lpKnwWKRwMRdSRcRtRpE1DUkEhIzu5hoMPmX7v67tOuRyUFdQyIigVOLQEQkcFNujKCxsdGXLFmSdhkiIlPKgw8+uMvdm4ptm3JBsGTJEtasGe5sQhERKcbMtgy3TV1DIiKBUxCIiAROQSAiErgpN0ZQTF9fH83NzfT09Bx85xJRXV3NggULqKioSLsUEZniSiIImpubqaurY8mSJeRNK1yy3J22tjaam5tZunRp2uWIyBSXWNeQmd1gZi1m9ugw283MrjWzjWb2yOAt+w5FT08PDQ0NQYQAgJnR0NAQVAtIRJKT5BjBt4lumjGc1UTzui8nuvXe157PNwslBAaFdrwikpzEuobc/XdmtmSEXc4FvuvRHBd/MrOZZnZYPJe6jKCnr5+93X3s7+7ji3duoKmuijeesICndnXy6DP7eOuqhTzR0sEvH93OwMBzpxCpqSznpCMauPeJVsyMNxw/n0yZ8bOHnqE7++xNtTJlZfztUY08tGUP+7v7JvoQJ0ymrIxTljeydute9nVl0y6npBw7fwZnHjOPzbs6+dmfn4E0p7Mx42+XN/KSJbMP2JTrH+C2R7azqbUDgJWHz8DdWb99/0RXeVBnrJjL3ywc7n5Dhy7NMYL5PPfWgM3xugOCwMwuIWo1sGjRosLNqWtra+OMM84AYMeOHWQyGZqaogv47r//fiorK4d97dNtXZSVwSc//D4++OGPctiSZVSXZ6itLqd5TzcDA051RRn7unOUZ4xcv5MbGKB/wGnvyfGVu7fiDlfe+tjQ/7Pv/HELf92xH3cobDgU/l/8tzs2DC3n7+sOX/qvxw9YX2pCOc6JNvh7dst7TuKbv9/Enet2pvrzdYdr73qCC162iH8+99jntKjfe9ND/HrdzqKvm2y/E3Pqq0suCIr9iIv+yeDu1wPXA6xatWrSzZLX0NDA2rVrAbjqqquora3lIx/5yHP2GbxJdFnZs71xHb059nZHf4X+67Vfp6W9hx37on7/MjMG4v9N+3sgU2YMuFFeZmTMWD6vlor2Gp767Gt4cMsefruhhRnTKmlt7+X7f9rCe087gktOXcbMac8NoS1tndyzoZUzj5nLgMPPH2pmwOH1x81nUcOzd0rc3Znl/z6yjZcsnc0L5tWP/w9tktjTmeW2v2znhEUzOebwGWmXUzK6sjnO/NLv+PAta9m+r4f3nLaMK1avSK2e7mw/n79zA9+69ynm1Vfz0NN7eenS2bz9ZYv5zV9beOdJi7n6nGMYcLjjsR2UGbxq5TwyZZMsCRKSZhA089x7yi4AtqVUSyI2btzI61//ek455RTuu+8+brvtNq6++moeeughuru7OfN1b+DiD3yMynLjzDNO458+/W+cdcqJLDh8Lu94199zz113Ul5VzZe/+QNefuxSKsszQBQq+X/RvHjxLF68eNbQ84+9+mjKhvkFXtwwnQtfPn3o+aWvWF50v9nTK3nHSUvG4acwuc2aXsk7Xrb44DvKmEyrLOd/n3c8F3zzPvoHnLetWnjwFyWopjLDJ16zgq27u/jSfz1B/4CzdutejpxTS/+A86qVczEzMgZnv3DEG9WVpDSD4FbgUjO7mejm3/vGY3zg6v98jHXbxrdvb+Xh9XzqdQe9r3lR69at43Nf/hr/6wvXMnt6JZdd8SmaGhqorTROOfV03vDGN3HKquOozJQxr76amsoM+/ft4zWvPoOvfvnzXPqBD3L3rT/i9OM+PvQ1DzZQPFwIiEykFy+exU0Xv5QNO9pZ1lSbdjmYGZe9cjl3rttJTUWG3Z1Z/vm2dWTKjBMWzTr4FyhhiQWBmf0QOB1oNLNmoptrVwC4+9eB24GzgY1AF/DupGpJ07JlR7DoBS9kx74eaqvKuemmH/DzH32fgf5+WnZsZ9tTT1B24vFUZMqoq44uDqupqWH16tUAnPTSl/D73/8+zUMQOWQnLJo1qT5kjzl8Bp987UqOnlvHx3/xFza3dfGCeXVMryqJS6oOWZJnDZ1/kO0OvG+8v++h/uU+nvoHfKh/v6qmBsPIDQzw+zWPcNMN/4df3HkP5TV1fPKD7yHXd+CZKvmDy5lMhlwud8A+InJoLjolugjzlvecxA33PsWJSw88kyg0mmtonLk7Lft72NLWRa4/Ortn5rQK6qoraN29l9raWl609DBmeAd/uOeutMsVCdbc+mquOHsFZ6yYm3YpqQu7PZSAwdP223v66IjPS2+sqyJjcMyLjuPoF6zghS98IcuWLePkk09OsVIRkciUu2fxqlWrvPDGNOvXr2fFivROTcvX2ZvjyfjCFIAZNRUsbojO0unt6ydTZpRnxqchNpmOW0QmNzN70N1XFdumFsE4y/YPAFBVnmF6VYZ59dVD26oqMmmVJSIyLAXBOMvmoiBYPqdWp3GKyJRQMoPFaXdx7evOkusfoK9/gPKyssRDIO3jFZHSURJBUF1dTVtbW2ofjtncAFvaumhp7yWbG6CyPNkf6+D9CKqrqw++s4jIQZRE19CCBQtobm6mtbV1wr93b18//e7s7uyjrcwwg4pMGX1tw080Nx4G71AmIvJ8lUQQVFRUpHKnri/euYFrf7OJaZUZurL9Q+svOXUZ/3S2zuYRkamhJIIgDZt3dXLtbzZSVV5GV7afo+bWsn1fD/XVFbw15Qm2RETGQkFwiO57qg2AD73qKD73y79yypFNfOhVy5lWWR7M1LUiUhoUBIfovk27aZheyUWnLGXb3m7esmrB0KRxIiJTiYLgEN331G5OXDqbikwZ15x7bNrliIgcspI4fXSibWrt4Jm93bxUsxaKSAlQEByCHz/YTJnB6gDvZCQipUdBMEa5/gF++mAzr3jBHObW64IuEZn6FARj9ERLBy3tvbz2RYenXYqIyLhQEIzRlrZOAI6ck/49WEVExoOCYIw2t3UBsLhhWsqViIiMDwXBGG1p66RheqWuGRCRkqHrCEZpf08f77vpIf66o12tAREpKWoRjNKDm/fw+yd20dreO3TrSRGRUqAgGKX1O/YPLTfWJjvFtIjIRFIQjNKGHe1Dy8uadMaQiJQOjRGM0oYd7fzd0U18/DUrWNaoIBCR0qEWwShkcwM82drB0fPqOXJOnW5KLyIlRUEwCo/vbKev31lxWF3apYiIjDsFwSj8et1OzOCkIxrSLkVEZNwpCEbhV4/u4CVLZjOnTpPMiUjpURAcxJa2TjbsbGf1sfPSLkVEJBEKgoP4a3za6IsXz0q5EhGRZCgIDmJwttHFs3U1sYiUJgXBQWxp62LmtApmTNMkcyJSmhINAjM7y8w2mNlGM7u8yPbFZnaXmT1iZveY2YIk6zkUW9q6NLeQiJS0xILAzDLAdcBqYCVwvpmtLNjt88B33f1FwDXAZ5Oq51Btbutk8WzNNioipSvJFsGJwEZ33+TuWeBm4NyCfVYCd8XLdxfZnqpsboBte7tZommnRaSEJRkE84Gtec+b43X5HgbeFC+/AagzswOu2jKzS8xsjZmtaW1tTaTYYpr3dDHgqGtIREpakkFQbEIeL3j+EeA0M/szcBrwDJA74EXu17v7Kndf1dTUNP6VDmPLbt2WUkRKX5KzjzYDC/OeLwC25e/g7tuANwKYWS3wJnffl2BNY7JlV3zqqFoEIlLCkmwRPAAsN7OlZlYJnAfcmr+DmTWa2WANVwA3JFjPmG1u62JaZUY3ohGRkpZYELh7DrgUuANYD9zi7o+Z2TVmdk682+nABjN7HJgLfCapeg7F07ujU0fNNO20iJSuRG9M4+63A7cXrLsyb/knwE+SrOH52NzWyVFzNPW0iJQ2XVk8jP4Bp3l3N4sbNVAsIqVNQTCM7fu6yfYPsEQDxSJS4hQEw9i8Kz51VFcVi0iJUxAMY2NLNP30kXN0o3oRKW0KgmFsbO2gvrqcprqqtEsREUmUgmAYT+zs4Mg5tTp1VERKnoJgGE+2drBcp46KSAAUBEXs6cyyqyOr8QERCYKCoIgnWjoADRSLSBgUBEWs2xbNe7fy8PqUKxERSZ6CoIhHt+2nsbaKOTpjSEQCoCAo4tFn9nHs/HqdMSQiQVAQFOjp62djSwfHHj4j7VJERCaEgqDA4zvbyQ04x87X+ICIhEFBUKB5Tzegu5KJSDgUBAXaOnoBaKzVQLGIhEFBUKCtMwvArGkVKVciIjIxFAQF2jqyzJpWQXlGPxoRCYM+7Qq0dfYye7puVi8i4VAQFGjryNKg8QERCYiCoEBbZ5bGWrUIRCQcCoICbR3qGhKRsCgI8uT6B9jT1UfDdHUNiUg4FAR59nT1AdCgriERCYiCIE9bZ3QxmVoEIhISBUGeto7oYjKNEYhISBQEeVrboxZBk+5DICIBURDk2bm/B4B5M6pTrkREZOIoCPLs2N/D9MoMtVXlaZciIjJhFAR5Wvb3MrderQERCYuCIM/O/T3Mqdf4gIiERUGQZ2d7D/PUIhCRwCgIYu7OTnUNiUiAEg0CMzvLzDaY2UYzu7zI9kVmdreZ/dnMHjGzs5OsZyR7u/rI5gaYoyAQkcAkFgRmlgGuA1YDK4HzzWxlwW6fAG5x9+OB84B/T6qeg9nZHp86qiAQkcAk2SI4Edjo7pvcPQvcDJxbsI8D9fHyDGBbgvWMaOf+6GIyDRaLSGiSDIL5wNa8583xunxXAReYWTNwO/D+Yl/IzC4xszVmtqa1tTWJWtnXHU04p3sVi0hokgwCK7LOC56fD3zb3RcAZwPfM7MDanL36919lbuvampqSqBU6Mn2A1BdkUnk64uITFZJBkEzsDDv+QIO7Pq5CLgFwN3/CFQDjQnWNKyubA6AaZW6qlhEwpJkEDwALDezpWZWSTQYfGvBPk8DZwCY2QqiIEim7+cguvsGAJhWqRaBiIQlsSBw9xxwKXAHsJ7o7KDHzOwaMzsn3u0fgIvN7GHgh8C73L2w+2hCdGdzmEFVuS6tEJGwJNoP4u63Ew0C56+7Mm95HXBykjWMVndfPzUVGcyKDW2IiJQu/fkb68pGQSAiEppRBYGZHWFmVfHy6Wb2ATObmWxpE6s720+NxgdEJECjbRH8FOg3syOBbwFLgR8kVlUKuvv6NVAsIkEabRAMxIO/bwC+7O4fAg5LrqyJp64hEQnVaIOgz8zOBy4EbovXldQluN196hoSkTCNNgjeDZwEfMbdnzKzpcD3kytr4nWrRSAigRrV6aPxaZ4fADCzWUCdu38uycImWlc2x7TKaWmXISIy4UZ71tA9ZlZvZrOBh4EbzeyLyZY2sXr6BjTPkIgEabRdQzPcfT/wRuBGd38x8Mrkypp4UYtAQSAi4RltEJSb2WHAW3l2sLik6PRREQnVaIPgGqI5g5509wfMbBnwRHJlTayBAVfXkIgEa7SDxT8Gfpz3fBPwpqSKmmg9ueheBGoRiEiIRjtYvMDMfm5mLWa208x+amYLki5uonTFN6XRdQQiEqLRdg3dSHQvgcOJbjf5n/G6ktA9GATqGhKRAI02CJrc/UZ3z8WPbwPJ3DMyBd19g11DujuZiIRntEGwy8wuMLNM/LgAaEuysIn0bNeQZuUWkfCM9pPvvxOdOroD2A68mWjaiZLwbNeQWgQiEp5RBYG7P+3u57h7k7vPcffXE11cVhK6+6Ib12uwWERC9Hz6Qj48blWkrL0nCoLaKgWBiITn+QRBydzcd+vuLgDmz9SkcyISnucTBD5uVaRsc1sXc+ur1DUkIkEacXTUzNop/oFvQE0iFaXg6bYuFs+ennYZIiKpGDEI3L1uogpJ0+a2Tk47qmQuixARGZPgT5zvyuZoae9lcYPGB0QkTMEHwdPxQPHiBnUNiUiYgg+CLW2DQaAWgYiEKfgg2N2ZBaCxtirlSkRE0hF8ELT39AFQV63pJUQkTMEHQUdPDjOYrplHRSRQwQfB/p4ctZXllJWVzIXSIiJjEnwQdPTm1C0kIkELPgjae/qoq65IuwwRkdQoCHpy1KpFICIBSzQIzOwsM9tgZhvN7PIi279kZmvjx+NmtjfJeopR15CIhC6xT0AzywDXAa8CmoEHzOxWd183uI+7fyhv//cDxydVz3Dae3K6qlhEgpZki+BEYKO7b3L3LHAzcO4I+58P/DDBeopq78lRW6UWgYiEK8kgmA9szXveHK87gJktBpYCvxlm+yVmtsbM1rS2to5rke09fdSra0hEApZkEBQ7MX+4m9mcB/zE3fuLbXT36919lbuvamoav+mis7kBenMDGiMQkaAlGQTNwMK85wuAbcPsex4pdAt19A7eq1hBICLhSjIIHgCWm9lSM6sk+rC/tXAnMzsamAX8McFainp2niFdRyAi4UosCNw9B1wK3AGsB25x98fM7BozOydv1/OBm919wu+B3N4TtQjUNSQiIUv0E9DdbwduL1h3ZcHzq5KsYSSDQaALykQkZEFfWTzYNVSvriERCVjQQbA/bhEoCEQkZEEHwTN7ujGDuTN0dzIRCVfQQfD07i7m1lVTVZ5JuxQRkdQEHQRb93SxcHZN2mWIiKQq6CBo3t3FwlnT0i5DRCRVwQZBNjfA9v09LJitIBCRsAUbBM/s7cYdFikIRCRwwQbB1t1dACycpTECEQlbsEGwfV83AIfPVBCISNiCDYKO3mjG6/oaXUwmImELNgi64imop1XqGgIRCVuwQdCRzVFZXkZFJtgfgYgIEHAQdPX2M12tARGRcIOgM5tjWqWmnxYRCTYIunr7mV6lFoGISLBB0JnNMV33KhYRCTcIurL9TFfXkIhIuEHQ2ZvTqaMiIoQcBOoaEhEBAg6Crt5+tQhERAg4CDqzOWrVIhARCTMI+gecnr4BXUcgIkKgQdCVjeYZ0nUEIiKBBkFnPPOoWgQiIqEGgVoEIiJDggyCLrUIRESGBBkEahGIiDwryCAYGixWi0BEJMwgGBwsVotARCTYIIhaBDVqEYiIhBkEu7uyAMyaphvXi4gkGgRmdpaZbTCzjWZ2+TD7vNXM1pnZY2b2gyTrGbSrPcv0yozOGhIRARL7JDSzDHAd8CqgGXjAzG5193V5+ywHrgBOdvc9ZjYnqXrytXb00lRXNRHfSkRk0kuyRXAisNHdN7l7FrgZOLdgn4uB69x9D4C7tyRYz5DW9h4FgYhILMkgmA9szXveHK/LdxRwlJn9wcz+ZGZnJVjPkF0dWRprFQQiIpBsEFiRdV7wvBxYDpwOnA9808xmHvCFzC4xszVmtqa1tfV5F9barq4hEZFBSQZBM7Aw7/kCYFuRff7D3fvc/SlgA1EwPIe7X+/uq9x9VVNT0/MqqjfXz77uPprUIhARAZINggeA5Wa21MwqgfOAWwv2+QXwdwBm1kjUVbQpwZpo64hOHW1Ui0BEBEgwCNw9B1wK3AGsB25x98fM7BozOyfe7Q6gzczWAXcDH3X3tqRqgqhbCFCLQEQkluiJ9O5+O3B7wbor85Yd+HD8mBBDQaAWgYgIEOCVxa0dURCoa0hEJBJcEOzujMYIGqZXplyJiMjkEFwQ7Ovuo6q8jOoKzTwqIgIBBsGeziyzpqk1ICIyKLgg2Nvdx0zNOioiMiS8IOjKKghERPIEGAR9zKxR15CIyKDggmBPVx+zpqtFICIyKKggcHf2dWeZoRaBiMiQoIKgM9tPX7/rFpUiInmCCoK98b2KNVgsIvKswIKgD4CZuo5ARGRImEFQoxaBiMigoIJgT9w1NEvzDImIDAkqCPZ2q0UgIlIoqCDY1d6LmVoEIiL5ggqClvYeGqZXUZEJ6rBFREYU1Cfizv29zK3XDWlERPIFFQQ79vUwt7467TJERCaVoIKgpb1HLQIRkQLBBEFf/wC7OrJqEYiIFAgmCFrbo5vWKwhERJ4rmCDYsb8HQF1DIiIFggmClqEgUItARCRfMEGwc7+6hkREigkmCA6bUc2ZK+cyWzOPiog8R3naBUyUM4+Zx5nHzEu7DBGRSSeYFoGIiBSnIBARCZyCQEQkcAoCEZHAKQhERAKnIBARCZyCQEQkcAoCEZHAmbunXcOYmFkrsOUQX94I7BrHctKkY5mcdCyTk44FFrt7U7ENUy4Ing8zW+Puq9KuYzzoWCYnHcvkpGMZmbqGREQCpyAQEQlcaEFwfdoFjCMdy+SkY5mcdCwjCGqMQEREDhRai0BERAooCEREAhdMEJjZWWa2wcw2mtnladczVma22cz+YmZrzWxNvG62mf3azJ6I/52Vdp3FmNkNZtZiZo/mrStau0Wujd+nR8zshPQqP9Awx3KVmT0TvzdrzezsvG1XxMeywcxenU7VBzKzhWZ2t5mtN7PHzOyyeP2Ue19GOJap+L5Um9n9ZvZwfCxXx+uXmtl98fvyIzOrjNdXxc83xtuXHNI3dveSfwAZ4ElgGVAJPAysTLuuMR7DZqCxYN2/ApfHy5cD/5J2ncPUfipwAvDowWoHzgZ+CRjwMuC+tOsfxbFcBXykyL4r49+1KmBp/DuYSfsY4toOA06Il+uAx+N6p9z7MsKxTMX3xYDaeLkCuC/+ed8CnBev/zrw3nj5fwJfj5fPA350KN83lBbBicBGd9/k7lngZuDclGsaD+cC34mXvwO8PsVahuXuvwN2F6wervZzge965E/ATDM7bGIqPbhhjmU45wI3u3uvuz8FbCT6XUydu29394fi5XZgPTCfKfi+jHAsw5nM74u7e0f8tCJ+OPAK4Cfx+sL3ZfD9+glwhpnZWL9vKEEwH9ia97yZkX9RJiMH7jSzB83sknjdXHffDtF/BmBOatWN3XC1T9X36tK4y+SGvC66KXEscXfC8UR/fU7p96XgWGAKvi9mljGztUAL8GuiFsted8/Fu+TXO3Qs8fZ9QMNYv2coQVAsIafaebMnu/sJwGrgfWZ2atoFJWQqvldfA44AjgO2A1+I10/6YzGzWuCnwAfdff9IuxZZN9mPZUq+L+7e7+7HAQuIWioriu0W/zsuxxJKEDQDC/OeLwC2pVTLIXH3bfG/LcDPiX5Bdg42z+N/W9KrcMyGq33KvVfuvjP+zzsAfINnuxkm9bGYWQXRB+dN7v6zePWUfF+KHctUfV8Gufte4B6iMYKZZlYeb8qvd+hY4u0zGH3X5ZBQguABYHk88l5JNKhya8o1jZqZTTezusFl4EzgUaJjuDDe7ULgP9Kp8JAMV/utwDvjs1ReBuwb7KqYrAr6yt9A9N5AdCznxWd2LAWWA/dPdH3FxP3I3wLWu/sX8zZNufdluGOZou9Lk5nNjJdrgFcSjXncDbw53q3wfRl8v94M/MbjkeMxSXuUfKIeRGc9PFgXyGAAAAIySURBVE7U3/bxtOsZY+3LiM5yeBh4bLB+or7Au4An4n9np13rMPX/kKhp3kf0F8xFw9VO1NS9Ln6f/gKsSrv+URzL9+JaH4n/Yx6Wt//H42PZAKxOu/68uk4h6kJ4BFgbP86eiu/LCMcyFd+XFwF/jmt+FLgyXr+MKKw2Aj8GquL11fHzjfH2ZYfyfTXFhIhI4ELpGhIRkWEoCEREAqcgEBEJnIJARCRwCgIRkcApCEQKmFl/3oyVa20cZ6s1syX5M5eKTAblB99FJDjdHl3iLxIEtQhERsmie0L8Szxf/P1mdmS8frGZ3RVPbnaXmS2K1881s5/Hc8s/bGYvj79Uxsy+Ec83f2d8BalIahQEIgeqKegaelvetv3ufiLwVeDL8bqvEk3R/CLgJuDaeP21wG/d/W+I7mHwWLx+OXCdux8D7AXelPDxiIxIVxaLFDCzDnevLbJ+M/AKd98UT3K2w90bzGwX0fQFffH67e7eaGatwAJ37837GkuAX7v78vj5PwIV7v7p5I9MpDi1CETGxodZHm6fYnrzlvvRWJ2kTEEgMjZvy/v3j/Hy/yOa0Rbg7cC98fJdwHth6GYj9RNVpMhY6C8RkQPVxHeIGvQrdx88hbTKzO4j+iPq/HjdB4AbzOyjQCvw7nj9ZcD1ZnYR0V/+7yWauVRkUtEYgcgoxWMEq9x9V9q1iIwndQ2JiAROLQIRkcCpRSAiEjgFgYhI4BQEIiKBUxCIiAROQSAiErj/D1pKeXeTOBKWAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "numEpochs=300\n",
    "train_model(autoencoder_model,x_train,y_train,epochs=numEpochs,batch_size=20)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "predictions => [ 1.04907846e+00  8.47971618e-01 -1.36626527e-01 -2.05323249e-02\n",
      "  9.61106271e-02  1.02915645e+00 -2.48379514e-01  9.02010620e-01\n",
      " -6.35435730e-02  1.06752664e-02  9.52737570e-01  1.01258588e+00\n",
      "  5.94216526e-01 -4.26058173e-02  8.15789461e-01  3.33873183e-02\n",
      "  2.77330875e-01 -2.73034573e-01  9.84727621e-01  1.43073902e-01\n",
      "  7.75110602e-01  2.29725510e-01  1.15857208e+00  4.58711922e-01\n",
      "  9.59521055e-01 -2.11697966e-02  4.16784286e-02  9.27541971e-01\n",
      "  1.07954693e+00 -2.37002969e-04  9.25771773e-01  8.81900430e-01\n",
      "  5.22562116e-02  1.01889706e+00  3.51743519e-01 -7.22063929e-02\n",
      "  1.53250039e-01  8.93365622e-01  9.02216434e-01  1.14968204e+00\n",
      "  9.51241136e-01  1.09081841e+00  1.44211531e-01  3.20255786e-01\n",
      " -1.90481246e-02  8.65473807e-01  7.10518003e-01  8.40582848e-01\n",
      "  6.86021566e-01  1.00215085e-01  1.05978109e-01  8.77248943e-01\n",
      "  6.32520139e-01 -1.22536466e-01  1.74554557e-01  1.07980222e-02\n",
      "  1.00210690e+00  9.10771489e-01  4.64332998e-01  9.65136647e-01\n",
      "  8.58361363e-01 -2.89446831e-01  2.58583009e-01  9.91460979e-01\n",
      "  1.06469572e+00 -6.27696663e-02  9.84996736e-01  6.16924524e-01\n",
      "  1.04589856e+00  3.90107661e-01  1.02696133e+00  5.93394220e-01\n",
      "  8.73144448e-01  9.04207468e-01  6.00795865e-01  4.48996603e-01\n",
      "  3.60454768e-02  4.75033522e-01 -2.34963298e-02  1.03839815e+00\n",
      "  1.41879022e-02  1.06145144e+00  1.07417083e+00  9.91800547e-01\n",
      "  4.78437781e-01  1.50338739e-01  1.00256956e+00  9.43817437e-01\n",
      "  9.67346132e-01  1.06562662e+00  8.88677776e-01  1.94549501e-01\n",
      "  1.04710495e+00  9.59238052e-01  3.25476348e-01  9.36538219e-01\n",
      "  1.03164542e+00  1.04575574e+00  1.32640138e-01 -3.64977717e-02\n",
      "  2.03522727e-01  2.80194640e-01  1.01223898e+00  2.25957215e-01\n",
      "  9.31357384e-01  1.02168417e+00  1.93482637e-02  5.29617071e-01\n",
      "  7.40970492e-01  5.13399482e-01  8.94673288e-01  8.83237720e-01\n",
      "  9.91800964e-01  4.35560793e-02  1.88081980e-01  5.34500331e-02\n",
      "  6.66801929e-01  1.02961755e+00 -1.29310936e-02  6.49900675e-01\n",
      "  2.36034542e-02  1.87699035e-01  1.51077315e-01  4.03109342e-01\n",
      "  8.14143866e-02  8.29916835e-01  9.85518634e-01  1.07526648e+00\n",
      "  1.02883494e+00  9.50184226e-01  1.00454247e+00  9.04185697e-02\n",
      "  1.01943350e+00  1.68350011e-01  7.61684000e-01  1.13554251e+00\n",
      "  1.00242412e+00  8.96065593e-01  9.99105126e-02  1.00040567e+00\n",
      "  8.28711033e-01 -2.22483709e-01  9.63955879e-01  4.65681136e-01\n",
      "  6.17962122e-01  9.24845099e-01  9.98564303e-01  1.01559138e+00\n",
      "  1.02526271e+00  1.06274021e+00 -2.46298462e-02  8.09580386e-01\n",
      " -1.48896277e-02  9.78438318e-01  1.79052234e-01  1.04131317e+00\n",
      "  8.84822905e-01  5.15870869e-01  3.28978837e-01  3.24295610e-01\n",
      "  4.36028689e-02  1.42744333e-01  6.87446177e-01  3.92017305e-01\n",
      "  9.96353328e-01  2.66154706e-01  1.52502626e-01  8.02561581e-01\n",
      "  8.77715707e-01 -4.42112535e-02  7.39346445e-01  5.84632635e-01\n",
      " -2.08141655e-02  2.95377523e-01  9.83316481e-01  4.96960878e-01\n",
      "  1.32187426e-01  5.29949844e-01  1.09581327e+00  9.32333887e-01\n",
      "  2.16071248e-01  7.29077876e-01  9.99849021e-01  5.60066879e-01\n",
      "  1.03321922e+00  5.46347022e-01  1.61701694e-01  8.65315795e-02\n",
      "  9.54082549e-01  3.33312452e-02  8.95916343e-01  8.64082396e-01\n",
      "  2.60475278e-03  1.01270461e+00  1.02155423e+00  7.13452041e-01\n",
      "  8.34349573e-01  9.53726232e-01  1.04530931e+00  4.78465438e-01]\n",
      "y_test => [1 1 0 0 0 1 0 1 0 0 0 0 1 0 1 0 0 0 1 0 1 0 1 1 0 0 0 1 1 0 1 1 1 0 0 0 0\n",
      " 1 1 1 1 1 0 0 0 1 1 0 0 1 1 0 1 0 0 0 1 1 0 1 0 0 0 1 1 1 0 1 1 0 1 0 0 1\n",
      " 0 0 0 1 0 1 0 1 1 1 1 1 0 1 1 1 1 0 1 1 0 1 1 1 0 0 0 0 1 0 0 1 0 1 0 1 0\n",
      " 1 1 0 1 0 1 0 0 1 0 0 0 0 0 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 0 0 1 1 0 1 1 1\n",
      " 1 1 0 1 0 1 0 1 1 1 0 0 0 0 0 1 1 0 0 1 1 0 0 0 0 0 1 0 0 1 1 1 0 0 1 1 1\n",
      " 0 0 0 1 1 1 1 0 1 1 0 1 1 1 0]\n"
     ]
    }
   ],
   "source": [
    "predictions=eval_model(autoencoder_model,x_test)\n",
    "print(\"predictions =>\",predictions.ravel())\n",
    "print(\"y_test =>\",y_test.ravel())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "200\n",
      "200\n"
     ]
    }
   ],
   "source": [
    "print(len(predictions.ravel()))\n",
    "print(len(y_test.ravel()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def convert_to_binary(pred_cls):\n",
    "    for i in range(len(pred_cls)):\n",
    "        if pred_cls[i]>=0.5:pred_cls[i]=1\n",
    "        else:pred_cls[i]=0\n",
    "    return pred_cls\n",
    "pred_classes=convert_to_binary(predictions.ravel())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[71 25]\n",
      " [12 92]]\n"
     ]
    }
   ],
   "source": [
    "cm1 = confusion_matrix(y_test.ravel(),pred_classes)\n",
    "print(cm1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "For 2D confusion matrix\n",
      "Confusion Matrix :\n",
      "[[71 25]\n",
      " [12 92]]\n",
      "Accuracy Score : 81.5\n",
      "Report : \n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.86      0.74      0.79        96\n",
      "           1       0.79      0.88      0.83       104\n",
      "\n",
      "    accuracy                           0.81       200\n",
      "   macro avg       0.82      0.81      0.81       200\n",
      "weighted avg       0.82      0.81      0.81       200\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import accuracy_score \n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "print(\"For 2D confusion matrix\")\n",
    "\n",
    "tn, fp, fn, tp = cm1.ravel()\n",
    "accuracy=(tp+tn)/(tp+tn+fp+fn)\n",
    "accuracy*=100\n",
    "recall=tp/(tp+fn)\n",
    "recall*=100\n",
    "precision=tp/(tp+fp)\n",
    "precision*=100\n",
    "f1_score=2*(recall*precision)/(recall+precision)\n",
    "\n",
    "#print(\"accuracy : \",accuracy)\n",
    "#print(\"recall : \",recall)\n",
    "#print(\"precision : \",precision)\n",
    "#print(\"f1_score : \",f1_score)\n",
    "\n",
    "actual=y_test.ravel()\n",
    "predicted=pred_classes\n",
    "results = confusion_matrix(actual, predicted) \n",
    "print('Confusion Matrix :')\n",
    "print(results) \n",
    "print('Accuracy Score :',100*accuracy_score(actual, predicted))  \n",
    "print('Report : ')\n",
    "print(classification_report(actual, predicted) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}