AML Project_V0.5_Autoencoders_June_Train_VG.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "ec8a2606",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "from datetime import date\n",
    "import datetime as dt\n",
    "import tensorflow as tf\n",
    "import keras\n",
    "from keras.layers import Input,Dropout,Dense,LSTM,TimeDistributed,RepeatVector\n",
    "from keras.models import Model\n",
    "from keras import regularizers\n",
    "from sklearn.preprocessing import StandardScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "71598afd",
   "metadata": {},
   "outputs": [],
   "source": [
    "class DataPreprocessor:\n",
    "    def __init__(self, data_path):\n",
    "        self.data_path = data_path\n",
    "        self.df = pd.read_csv(self.data_path)\n",
    "        self.df = self.df.drop(self.df.columns[0], axis=1)\n",
    "        self.df['timestamp'] = pd.to_datetime(self.df['timestamp'])\n",
    "        self.train_data = None\n",
    "        self.test_data = None\n",
    "        self.analog_train = None\n",
    "        self.analog_test = None\n",
    "        self.digital_train = None\n",
    "        self.digital_test = None\n",
    "\n",
    "    def preprocessing_df(self):\n",
    "        # create a new column where 1 day before the timeframe as well as the timeframe indicated in the paper is labeled as unhealthy data i.e. 1\n",
    "        # healthy data is set as 0\n",
    "        self.df['is_anomaly'] = np.where(\n",
    "            ((self.df['timestamp'] >= \"2020-04-11 11:50:00\") & (self.df['timestamp'] <= \"2020-04-12 23:30:00\")) |\n",
    "            ((self.df['timestamp'] >= \"2020-04-17 00:00:00\") & (self.df['timestamp'] <= \"2020-04-19 01:30:00\")) |\n",
    "            ((self.df['timestamp'] >= \"2020-04-28 03:20:00\") & (self.df['timestamp'] <= \"2020-04-29 22:20:00\")) |\n",
    "            \n",
    "            \n",
    "            ((self.df['timestamp'] >= \"2020-05-12 14:00:00\") & (self.df['timestamp'] <= \"2020-05-13 23:59:00\")) |\n",
    "            ((self.df['timestamp'] >= \"2020-05-17 05:00:00\") & (self.df['timestamp'] <= \"2020-05-20 20:00:00\")) |\n",
    "            \n",
    "            \n",
    "            ((self.df['timestamp'] >= \"2020-05-28 23:30:00\") & (self.df['timestamp'] <= \"2020-05-30 06:00:00\")) |\n",
    "            \n",
    "            ((self.df['timestamp'] >= \"2020-05-31 15:00:00\") & (self.df['timestamp'] <= \"2020-06-01 15:40:00\")) |\n",
    "            ((self.df['timestamp'] >= \"2020-06-02 10:00:00\") & (self.df['timestamp'] <= \"2020-06-03 11:00:00\")) |\n",
    "            ((self.df['timestamp'] >= \"2020-06-04 10:00:00\") & (self.df['timestamp'] <= \"2020-06-07 14:30:00\")) |\n",
    "            \n",
    "            ((self.df['timestamp'] >= \"2020-07-07 17:30:00\") & (self.df['timestamp'] <= \"2020-07-08 19:00:00\")) |\n",
    "            ((self.df['timestamp'] >= \"2020-07-14 14:30:00\") & (self.df['timestamp'] <= \"2020-07-15 19:00:00\")) |\n",
    "            ((self.df['timestamp'] >= \"2020-07-16 04:30:00\") & (self.df['timestamp'] <= \"2020-07-17 05:30:00\"))\n",
    "            ,\n",
    "            1, 0\n",
    "        )\n",
    "        self.data_path = \"dataset_train_processed.csv\" # change this path accordingly when you want to change the file location\n",
    "        self.df.to_csv(self.data_path)\n",
    "        self.df = pd.read_csv(self.data_path)\n",
    "\n",
    "    def preprocessing_autoencoder(self):\n",
    "        # self.train_data = self.df[\n",
    "        #     (self.df['timestamp'] >= \"2022-01-01 06:00:00\") & (self.df['timestamp'] <= \"2022-02-28 02:00:00\")]\n",
    "        # self.test_data = self.df[\n",
    "        #     (self.df['timestamp'] >= \"2022-02-28 06:00:00\")]\n",
    "        self.train_data = self.df[\n",
    "            (self.df['timestamp'] >= \"2020-02-01 00:00:00\") & (self.df['timestamp'] < \"2020-06-30 11:59:59\") & \n",
    "            (self.df['is_anomaly'] == 0)]\n",
    "        self.test_data = self.df[\n",
    "            (self.df['timestamp'] >= \"2020-07-01 00:00:00\") & (self.df['timestamp'] < \"2020-07-31 11:59:59\")]\n",
    "        self.train_data.drop(self.train_data.columns[0], axis=1,\n",
    "                             inplace=True)  # there is an additional unnecessary column created at index 0 from preprocessing.df that needs to be removed if preprocessing_df is called\n",
    "        self.test_data.drop(self.test_data.columns[0], axis=1, inplace=True)\n",
    "        scaler = StandardScaler() # standardising only the analog data, leaving the digital data as is\n",
    "        self.analog_train = pd.DataFrame(scaler.fit_transform(self.train_data.iloc[:, 1:8]))\n",
    "        self.digital_train = self.train_data.iloc[:, 8:16]\n",
    "        print(self.test_data)\n",
    "        print(max(self.df['timestamp']))\n",
    "        self.analog_test = pd.DataFrame(scaler.transform(self.test_data.iloc[:, 1:8]))\n",
    "        self.digital_test = self.test_data.iloc[:, 8:16]\n",
    "\n",
    "    def preprocessing_arima(self):\n",
    "        df = self.df\n",
    "        columns = [\"timestamp\", \"TP2\"]\n",
    "        df = df[columns]\n",
    "        df['timestamp'] = pd.to_datetime(df['timestamp'])\n",
    "        df = df.resample(\"5T\", on=\"timestamp\").mean() # data can also be resampled to other timeframes: refer to pandas documentation for more info\n",
    "        df = df.fillna(method=\"ffill\")\n",
    "        self.train_data = df[(df.index >= pd.to_datetime(\"2020-02-01 00:00:00\")) & (df.index <= pd.to_datetime(\"2020-04-30 11:59:59\"))]\n",
    "        self.test_data = df[(df.index >= pd.to_datetime(\"2020-05-01 00:00:00\")) & (df.index <= pd.to_datetime(\"2020-05-04 11:59:59\"))] # change accordingly depending on your forecast timeframe\n",
    "\n",
    "    def preprocessing_ma(self):\n",
    "        columns = [\"timestamp\", \"TP2\"]\n",
    "        self.df = self.df[columns]\n",
    "        self.train_data = self.df[\n",
    "            (self.df['timestamp'] >= \"2020-02-01 00:00:00\") & (self.df['timestamp'] <= \"2020-04-30 11:59:59\")]\n",
    "        self.test_data = self.df[\n",
    "            (self.df['timestamp'] >= \"2020-05-01 00:00:00\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "7cbdd869",
   "metadata": {},
   "outputs": [],
   "source": [
    "def main():\n",
    "    preprocessor = DataPreprocessor(\"MetroPT3(AirCompressor).CSV\") # change this to the file path of the train dataset\n",
    "    preprocessor.preprocessing_df()\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "4a521880",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import matplotlib.pyplot as plt\n",
    "import pickle\n",
    "import keras.backend as K\n",
    "from tensorflow import keras\n",
    "from keras.models import Model, save_model\n",
    "from keras.layers import Conv1D, Conv1DTranspose, Lambda, Reshape, GlobalAveragePooling1D\n",
    "from keras.callbacks import EarlyStopping\n",
    "from tensorflow.python.framework.ops import disable_eager_execution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "3e57bc67",
   "metadata": {},
   "outputs": [],
   "source": [
    "disable_eager_execution() # variational autoencoder will not work without this\n",
    "# an alternative is to replace all instances of keras with tf.keras"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "4ae0c884",
   "metadata": {},
   "outputs": [],
   "source": [
    "class AutoencoderModels:\n",
    "\n",
    "    def __init__(self, train_data, test_data):\n",
    "        self.train_data = train_data\n",
    "        self.test_data = test_data\n",
    "        self.model = None\n",
    "        self.hist = None\n",
    "\n",
    "    def leaky_relu(self, x):  # to be used as the activation function of the output layer for the analog models; greatly reduces the loss value\n",
    "        return tf.nn.leaky_relu(x, alpha=0.01)\n",
    "\n",
    "    def sae_analog(self):  # both analog and digital sensors have 8 features/columns\n",
    "        # Encoder portion\n",
    "        input_layer = keras.Input(shape=(1, self.train_data.shape[1])) # shape of input is (no. of timesteps = 1, no. of features = 8) - same for all models\n",
    "        hidden_1 = Conv1D(filters=32, padding=\"same\", kernel_size=3, activation='relu')(input_layer)\n",
    "        hidden_2 = Conv1D(filters=16, padding=\"same\", kernel_size=3, activation='relu')(hidden_1)\n",
    "        bottleneck = Conv1D(filters=6, padding=\"same\", kernel_size=3, activation=\"relu\", activity_regularizer=keras.regularizers.l1_l2(l1=0.005, l2=0.1))(hidden_2)\n",
    "        #  activity_regularizer=keras.regularizers.l1_l2(l1=0.005, l2=0.1)\n",
    "        # Decoder portion\n",
    "        hidden_3 = Conv1DTranspose(filters=16, padding=\"same\", kernel_size=3, activation='relu')(bottleneck)\n",
    "        hidden_4 = Conv1DTranspose(filters=32, padding=\"same\", kernel_size=3, activation='relu')(hidden_3)\n",
    "        output_layer = Conv1DTranspose(filters=self.train_data.shape[1], padding=\"same\", kernel_size=3,\n",
    "                              activation=self.leaky_relu)(\n",
    "            hidden_4)\n",
    "        model = Model(inputs=input_layer, outputs=output_layer)\n",
    "        model.compile(optimizer='adam', loss='mse') # loss function can also be set to RMSE, MAE etc. for parameter tuning\n",
    "        callback = EarlyStopping(monitor=\"val_loss\", patience=5, restore_best_weights=True)\n",
    "        hist = model.fit(np.expand_dims(self.train_data, 1), np.expand_dims(self.train_data, 1), validation_split=0.2,\n",
    "                         epochs=50, callbacks=[callback], batch_size=60)\n",
    "        save_model(model, \"sae_analog.h5\")\n",
    "        self.model = model\n",
    "        self.hist = hist\n",
    "\n",
    "    def sae_digital(self):\n",
    "        # Encoder portion\n",
    "        input_layer = keras.Input(shape=(1, self.train_data.shape[1]))\n",
    "        hidden_1 = Conv1D(filters=32, padding=\"same\", kernel_size=3, activation='relu')(input_layer)\n",
    "        hidden_2 = Conv1D(filters=16, padding=\"same\", kernel_size=3, activation='relu')(hidden_1)\n",
    "        bottleneck = Conv1D(filters=6, padding=\"same\", kernel_size=3, activation=\"relu\")(hidden_2)\n",
    "\n",
    "        # Decoder Portion\n",
    "        hidden_3 = Conv1DTranspose(filters=16, padding=\"same\", kernel_size=3, activation='relu')(bottleneck)\n",
    "        hidden_4 = Conv1DTranspose(filters=32, padding=\"same\", kernel_size=3, activation='relu')(hidden_3)\n",
    "        output_layer = Conv1DTranspose(filters=self.train_data.shape[1], padding=\"same\", kernel_size=3, activation=\"sigmoid\")(\n",
    "            hidden_4)\n",
    "        model = Model(inputs=input_layer, outputs=output_layer)\n",
    "        model.compile(optimizer='adam', loss='binary_crossentropy')\n",
    "        callback = EarlyStopping(monitor=\"val_loss\", patience=5, restore_best_weights=True)\n",
    "        hist = model.fit(np.expand_dims(self.train_data, 1), np.expand_dims(self.train_data, 1), validation_split=0.2,\n",
    "                         epochs=50, callbacks=[callback], batch_size=60)\n",
    "        save_model(model, \"sae_digital.h5\")\n",
    "        self.model = model\n",
    "        self.hist = hist\n",
    "\n",
    "    def vae_analog(self):\n",
    "\n",
    "        def sampling(args):  # reparameterization trick used in VAE to sample data points from the latent space\n",
    "            z_mean, z_log_var = args  # log variance instead of variance is better for data that is not assumed to have a normal distribution\n",
    "            batch_size = K.shape(z_mean)[0]\n",
    "            dim = K.int_shape(z_mean)[1]\n",
    "            epsilon = K.random_normal(shape=(batch_size, dim)) # is a normal distribution with mean 0, variance 1\n",
    "            return z_mean + K.exp(0.5 * z_log_var) * epsilon\n",
    "\n",
    "        def vae_loss(x_actual, x_decoded):\n",
    "            reconstruction_loss = K.mean(K.square(x_actual - x_decoded)) # measures the ability of the VAE to reconstruct the input data accurately\n",
    "            kl_loss = -0.5 * K.sum(1 + z_log_var - K.square(z_mean) - K.exp(z_log_var), axis=-1) # measures the difference between the learned latent distribution and the distribution of the data\n",
    "            return reconstruction_loss + kl_loss\n",
    "\n",
    "        # Encoder portion\n",
    "        input_layer = keras.Input(shape=(1, self.train_data.shape[1]))\n",
    "        hidden_1 = Conv1D(filters=32, padding=\"same\", kernel_size=3, activation='relu')(input_layer)\n",
    "        hidden_2 = Conv1D(filters=16, padding=\"same\", kernel_size=3, activation='relu')(hidden_1)\n",
    "        z_mean = Conv1D(filters=6, padding=\"same\", kernel_size=3, activation=\"linear\")(hidden_2)\n",
    "        z_log_var = Conv1D(filters=6, padding=\"same\", kernel_size=3, activation=\"linear\")(hidden_2)\n",
    "        z = Lambda(sampling)([z_mean, z_log_var])  # this layer is the lower-dimensional representation of the input data learned from the sampling method i.e. the latent space\n",
    "        # the combination of z_mean z_log_var and the sampling layer can be thought of as similar to the bottleneck layer in a traditional autoencoder\n",
    "\n",
    "        # Decoder portion\n",
    "        hidden_3 = Conv1DTranspose(filters=16, padding=\"same\", kernel_size=3, activation='relu')(z)\n",
    "        hidden_4 = Conv1DTranspose(filters=32, padding=\"same\", kernel_size=3, activation='relu')(hidden_3)\n",
    "        output_layer = Conv1DTranspose(filters=self.train_data.shape[1], padding=\"same\", kernel_size=3,\n",
    "                              activation=self.leaky_relu)(hidden_4)\n",
    "        output_layer = GlobalAveragePooling1D()(output_layer)\n",
    "        output_layer = Reshape((1, self.train_data.shape[1]))(output_layer)\n",
    "        model = Model(inputs=input_layer, outputs=output_layer)\n",
    "        model.compile(optimizer='adam', loss=vae_loss)\n",
    "        callback = EarlyStopping(monitor=\"val_loss\", patience=5, restore_best_weights=True)\n",
    "        hist = model.fit(np.expand_dims(self.train_data, 1), np.expand_dims(self.train_data, 1), validation_split=0.2,\n",
    "                         epochs=50, callbacks=[callback], batch_size=60)\n",
    "        save_model(model,\"vae_analog.h5\")\n",
    "        self.model = model\n",
    "        self.hist = hist\n",
    "\n",
    "    def vae_digital(self): # exact same architecture as vae_analog\n",
    "\n",
    "        def sampling(args):\n",
    "            z_mean, z_log_var = args\n",
    "            batch_size = K.shape(z_mean)[0]\n",
    "            dim = K.int_shape(z_mean)[1]\n",
    "            epsilon = K.random_normal(shape=(batch_size, dim))\n",
    "            return z_mean + K.exp(0.5 * z_log_var) * epsilon\n",
    "\n",
    "        def vae_loss(x_actual, x_decoded):\n",
    "            reconstruction_loss = K.mean(K.square(x_actual - x_decoded)) # measures the ability of the VAE to reconstruct the input data accurately\n",
    "            kl_loss = -0.5 * K.sum(1 + z_log_var - K.square(z_mean) - K.exp(z_log_var), axis=-1) # measures the difference between the learned latent distribution and the distribution of the data\n",
    "            return reconstruction_loss + kl_loss\n",
    "\n",
    "        # Encoder portion\n",
    "        input_layer = keras.Input(shape=(1, self.train_data.shape[1]))\n",
    "        hidden_1 = Conv1D(filters=32, padding=\"same\", kernel_size=3, activation='relu')(input_layer)\n",
    "        hidden_2 = Conv1D(filters=16, padding=\"same\", kernel_size=3, activation='relu')(hidden_1)\n",
    "        z_mean = Conv1D(filters=6, padding=\"same\", kernel_size=3, activation=\"linear\")(hidden_2)\n",
    "        z_log_var = Conv1D(filters=6, padding=\"same\", kernel_size=3, activation=\"linear\")(hidden_2)\n",
    "        z = Lambda(sampling)([z_mean, z_log_var])\n",
    "\n",
    "        # Decoder portion\n",
    "        hidden_3 = Conv1DTranspose(filters=16, padding=\"same\", kernel_size=3, activation='relu')(z)\n",
    "        hidden_4 = Conv1DTranspose(filters=32, padding=\"same\", kernel_size=3, activation='relu')(hidden_3)\n",
    "        output_layer = Conv1D(filters=self.train_data.shape[1], padding=\"same\", kernel_size=3,\n",
    "                              activation=\"sigmoid\")(hidden_4)\n",
    "        output_layer = GlobalAveragePooling1D()(output_layer)\n",
    "        output_layer = Reshape((1, self.train_data.shape[1]))(output_layer)\n",
    "        model = Model(inputs=input_layer, outputs=output_layer)\n",
    "        model.compile(optimizer='adam', loss=vae_loss)\n",
    "        callback = EarlyStopping(monitor=\"val_loss\", patience=5, restore_best_weights=True)\n",
    "        hist = model.fit(np.expand_dims(self.train_data, 1), np.expand_dims(self.train_data, 1), validation_split=0.2,\n",
    "                         epochs=50, callbacks=[callback], batch_size=60)\n",
    "        save_model(model, \"vae_digital.h5\")\n",
    "        self.model = model\n",
    "        self.hist = hist\n",
    "\n",
    "    def autoencoder_predict_digital(self):\n",
    "        train_pred = self.model.predict(np.expand_dims(self.train_data, 1))\n",
    "        print(\"Shape of train_pred before reshaping:\", train_pred.shape)\n",
    "        \n",
    "        train_pred = np.reshape(train_pred, (len(train_pred), 8))\n",
    "    \n",
    "        test_pred = self.model.predict(np.expand_dims(self.test_data, 1))\n",
    "        print(\"Shape of test_pred before reshaping:\", test_pred.shape)\n",
    "        test_pred = np.reshape(test_pred, (len(test_pred), 8))\n",
    "        return train_pred, test_pred\n",
    "    \n",
    "    def autoencoder_predict_analog(self):\n",
    "        train_pred = self.model.predict(np.expand_dims(self.train_data, 1))\n",
    "        print(\"Shape of train_pred before reshaping:\", train_pred.shape)\n",
    "        \n",
    "        train_pred = np.reshape(train_pred, (len(train_pred), 7))\n",
    "    \n",
    "        test_pred = self.model.predict(np.expand_dims(self.test_data, 1))\n",
    "        print(\"Shape of test_pred before reshaping:\", test_pred.shape)\n",
    "        test_pred = np.reshape(test_pred, (len(test_pred), 7))\n",
    "        return train_pred, test_pred\n",
    "\n",
    "    def plot_losses(self):\n",
    "        plt.plot(self.hist.history[\"loss\"])\n",
    "        plt.plot(self.hist.history[\"val_loss\"])\n",
    "        plt.title(\"Training/Validation Loss\")\n",
    "        plt.xlabel(\"epoch\")\n",
    "        plt.ylabel(\"loss\")\n",
    "        plt.legend([\"training_loss\", \"validation_loss\"])\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "e0da8437",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:53: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.train_data.drop(self.train_data.columns[0], axis=1,\n",
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:55: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data.drop(self.test_data.columns[0], axis=1, inplace=True)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "           TP2    TP3     H1  DV_pressure  Reservoirs  Oil_temperature  \\\n",
      "1073346 -0.012  8.822  8.810       -0.022       8.822           61.825   \n",
      "1073347 -0.012  8.804  8.792       -0.020       8.804           61.625   \n",
      "1073348 -0.012  8.786  8.774       -0.020       8.786           61.575   \n",
      "1073349 -0.012  8.770  8.758       -0.022       8.768           61.450   \n",
      "1073350 -0.012  8.752  8.740       -0.022       8.752           61.350   \n",
      "...        ...    ...    ...          ...         ...              ...   \n",
      "1291621 -0.012  8.188  8.178       -0.018       8.192           61.200   \n",
      "1291622 -0.014  8.174  8.164       -0.018       8.180           61.175   \n",
      "1291623 -0.012  8.162  8.152       -0.018       8.166           61.075   \n",
      "1291624 -0.014  8.148  8.138       -0.016       8.154           61.050   \n",
      "1291625 -0.014  8.136  2.940       -0.018       8.142           60.975   \n",
      "\n",
      "         Motor_current  COMP  DV_eletric  Towers  MPG  LPS  Pressure_switch  \\\n",
      "1073346         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073347         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073348         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073349         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073350         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "...                ...   ...         ...     ...  ...  ...              ...   \n",
      "1291621         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291622         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291623         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291624         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291625         5.3525   0.0         1.0     1.0  0.0  0.0              1.0   \n",
      "\n",
      "         Oil_level  Caudal_impulses  is_anomaly  \n",
      "1073346        1.0              1.0           0  \n",
      "1073347        1.0              1.0           0  \n",
      "1073348        1.0              1.0           0  \n",
      "1073349        1.0              1.0           0  \n",
      "1073350        1.0              1.0           0  \n",
      "...            ...              ...         ...  \n",
      "1291621        1.0              1.0           0  \n",
      "1291622        1.0              1.0           0  \n",
      "1291623        1.0              1.0           0  \n",
      "1291624        1.0              1.0           0  \n",
      "1291625        1.0              1.0           0  \n",
      "\n",
      "[218280 rows x 16 columns]\n",
      "2020-09-01 03:59:50\n",
      "Train on 752729 samples, validate on 188183 samples\n",
      "Epoch 1/50\n",
      "752640/752729 [============================>.] - ETA: 0s - loss: 0.2518"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\anaconda3\\lib\\site-packages\\keras\\src\\engine\\training_v1.py:2335: UserWarning: `Model.state_updates` will be removed in a future version. This property should not be used in TensorFlow 2.0, as `updates` are applied automatically.\n",
      "  updates = self.state_updates\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "752729/752729 [==============================] - 153s 203us/sample - loss: 0.2518 - val_loss: 0.0945\n",
      "Epoch 2/50\n",
      "752729/752729 [==============================] - 142s 189us/sample - loss: 0.0526 - val_loss: 0.0535\n",
      "Epoch 3/50\n",
      "752729/752729 [==============================] - 149s 198us/sample - loss: 0.0238 - val_loss: 0.0223\n",
      "Epoch 4/50\n",
      "752729/752729 [==============================] - 164s 218us/sample - loss: 0.0138 - val_loss: 0.0125\n",
      "Epoch 5/50\n",
      "752729/752729 [==============================] - 158s 210us/sample - loss: 0.0088 - val_loss: 0.0087\n",
      "Epoch 6/50\n",
      "752729/752729 [==============================] - 139s 184us/sample - loss: 0.0062 - val_loss: 0.0072\n",
      "Epoch 7/50\n",
      "752729/752729 [==============================] - 134s 178us/sample - loss: 0.0051 - val_loss: 0.0056\n",
      "Epoch 8/50\n",
      "752729/752729 [==============================] - 150s 199us/sample - loss: 0.0050 - val_loss: 0.0073\n",
      "Epoch 9/50\n",
      "752729/752729 [==============================] - 174s 231us/sample - loss: 0.0042 - val_loss: 0.0039\n",
      "Epoch 10/50\n",
      "752729/752729 [==============================] - 200s 265us/sample - loss: 0.0038 - val_loss: 0.0091\n",
      "Epoch 11/50\n",
      "752729/752729 [==============================] - 236s 313us/sample - loss: 0.0035 - val_loss: 0.0039\n",
      "Epoch 12/50\n",
      "752729/752729 [==============================] - 233s 310us/sample - loss: 0.0034 - val_loss: 0.0048\n",
      "Epoch 13/50\n",
      "752729/752729 [==============================] - 207s 275us/sample - loss: 0.0033 - val_loss: 0.0030\n",
      "Epoch 14/50\n",
      "752729/752729 [==============================] - 204s 271us/sample - loss: 0.0031 - val_loss: 0.0032\n",
      "Epoch 15/50\n",
      "752729/752729 [==============================] - 194s 257us/sample - loss: 0.0029 - val_loss: 0.0025\n",
      "Epoch 16/50\n",
      "752729/752729 [==============================] - 203s 270us/sample - loss: 0.0027 - val_loss: 0.0028\n",
      "Epoch 17/50\n",
      "752729/752729 [==============================] - 140s 186us/sample - loss: 0.0026 - val_loss: 0.0024\n",
      "Epoch 18/50\n",
      "752729/752729 [==============================] - 181s 240us/sample - loss: 0.0026 - val_loss: 0.0039\n",
      "Epoch 19/50\n",
      "752729/752729 [==============================] - 173s 230us/sample - loss: 0.0024 - val_loss: 0.0023\n",
      "Epoch 20/50\n",
      "752729/752729 [==============================] - 174s 232us/sample - loss: 0.0024 - val_loss: 0.0023\n",
      "Epoch 21/50\n",
      "752729/752729 [==============================] - 159s 211us/sample - loss: 0.0023 - val_loss: 0.0056\n",
      "Epoch 22/50\n",
      "752729/752729 [==============================] - 155s 206us/sample - loss: 0.0024 - val_loss: 0.0027\n",
      "Epoch 23/50\n",
      "752729/752729 [==============================] - 156s 208us/sample - loss: 0.0023 - val_loss: 0.0037\n",
      "Epoch 24/50\n",
      "752729/752729 [==============================] - 155s 206us/sample - loss: 0.0022 - val_loss: 0.0016\n",
      "Epoch 25/50\n",
      "752729/752729 [==============================] - 152s 202us/sample - loss: 0.0022 - val_loss: 0.0017\n",
      "Epoch 26/50\n",
      "752729/752729 [==============================] - 124s 165us/sample - loss: 0.0022 - val_loss: 0.0014\n",
      "Epoch 27/50\n",
      "752729/752729 [==============================] - 129s 172us/sample - loss: 0.0020 - val_loss: 0.0023\n",
      "Epoch 28/50\n",
      "752729/752729 [==============================] - 120s 160us/sample - loss: 0.0022 - val_loss: 0.0024\n",
      "Epoch 29/50\n",
      "752729/752729 [==============================] - 127s 169us/sample - loss: 0.0021 - val_loss: 0.0019\n",
      "Epoch 30/50\n",
      "752729/752729 [==============================] - 126s 167us/sample - loss: 0.0020 - val_loss: 0.0023\n",
      "Epoch 31/50\n",
      "752729/752729 [==============================] - 120s 159us/sample - loss: 0.0020 - val_loss: 0.0025\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\1723690628.py:90: UserWarning: You are saving your model as an HDF5 file via `model.save()`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')`.\n",
      "  save_model(model,\"vae_analog.h5\")\n",
      "C:\\Users\\vgupt\\AppData\\Local\\anaconda3\\lib\\site-packages\\keras\\src\\engine\\training_v1.py:2359: UserWarning: `Model.state_updates` will be removed in a future version. This property should not be used in TensorFlow 2.0, as `updates` are applied automatically.\n",
      "  updates=self.state_updates,\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of train_pred before reshaping: (940912, 1, 7)\n",
      "Shape of test_pred before reshaping: (218280, 1, 7)\n",
      "Model: \"model_3\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                Output Shape                 Param #   Connected to                  \n",
      "==================================================================================================\n",
      " input_4 (InputLayer)        [(None, 1, 7)]               0         []                            \n",
      "                                                                                                  \n",
      " conv1d_12 (Conv1D)          (None, 1, 32)                704       ['input_4[0][0]']             \n",
      "                                                                                                  \n",
      " conv1d_13 (Conv1D)          (None, 1, 16)                1552      ['conv1d_12[0][0]']           \n",
      "                                                                                                  \n",
      " conv1d_14 (Conv1D)          (None, 1, 6)                 294       ['conv1d_13[0][0]']           \n",
      "                                                                                                  \n",
      " conv1d_15 (Conv1D)          (None, 1, 6)                 294       ['conv1d_13[0][0]']           \n",
      "                                                                                                  \n",
      " lambda_2 (Lambda)           (None, None, 6)              0         ['conv1d_14[0][0]',           \n",
      "                                                                     'conv1d_15[0][0]']           \n",
      "                                                                                                  \n",
      " conv1d_transpose_8 (Conv1D  (None, None, 16)             304       ['lambda_2[0][0]']            \n",
      " Transpose)                                                                                       \n",
      "                                                                                                  \n",
      " conv1d_transpose_9 (Conv1D  (None, None, 32)             1568      ['conv1d_transpose_8[0][0]']  \n",
      " Transpose)                                                                                       \n",
      "                                                                                                  \n",
      " conv1d_transpose_10 (Conv1  (None, None, 7)              679       ['conv1d_transpose_9[0][0]']  \n",
      " DTranspose)                                                                                      \n",
      "                                                                                                  \n",
      " global_average_pooling1d_2  (None, 7)                    0         ['conv1d_transpose_10[0][0]'] \n",
      "  (GlobalAveragePooling1D)                                                                        \n",
      "                                                                                                  \n",
      " reshape_2 (Reshape)         (None, 1, 7)                 0         ['global_average_pooling1d_2[0\n",
      "                                                                    ][0]']                        \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 5395 (21.07 KB)\n",
      "Trainable params: 5395 (21.07 KB)\n",
      "Non-trainable params: 0 (0.00 Byte)\n",
      "__________________________________________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def main():\n",
    "    preprocessor = DataPreprocessor(\"dataset_train_processed.csv\") # change according to where your data is stored\n",
    "    \n",
    "    preprocessor.preprocessing_autoencoder()\n",
    "    # autoencoder = AutoencoderModels(preprocessor.digital_train, preprocessor.digital_test) # comment out depending on whether digital/analog data is used\n",
    "    autoencoder = AutoencoderModels(preprocessor.analog_train, preprocessor.analog_test)\n",
    "    # autoencoder.sae_analog() # comment out according to which method you call from the AutoencoderModels class\n",
    "    # autoencoder.vae_digital()\n",
    "    autoencoder.vae_analog()\n",
    "    # autoencoder.vae_digital()\n",
    "    train_pred, test_pred = autoencoder.autoencoder_predict_analog()\n",
    "    with open(\"vae_analog_train_pred\", \"wb\") as file: # name of file path can change depending on where you want to save it\n",
    "        pickle.dump(train_pred, file)\n",
    "    with open(\"vae_analog_test_pred\", \"wb\") as file: # similarly, this file path can be changed\n",
    "        pickle.dump(test_pred, file)\n",
    "    print(autoencoder.model.summary())\n",
    "    autoencoder.plot_losses()\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "feb1a7d2",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:53: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.train_data.drop(self.train_data.columns[0], axis=1,\n",
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:55: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data.drop(self.test_data.columns[0], axis=1, inplace=True)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "           TP2    TP3     H1  DV_pressure  Reservoirs  Oil_temperature  \\\n",
      "1073346 -0.012  8.822  8.810       -0.022       8.822           61.825   \n",
      "1073347 -0.012  8.804  8.792       -0.020       8.804           61.625   \n",
      "1073348 -0.012  8.786  8.774       -0.020       8.786           61.575   \n",
      "1073349 -0.012  8.770  8.758       -0.022       8.768           61.450   \n",
      "1073350 -0.012  8.752  8.740       -0.022       8.752           61.350   \n",
      "...        ...    ...    ...          ...         ...              ...   \n",
      "1291621 -0.012  8.188  8.178       -0.018       8.192           61.200   \n",
      "1291622 -0.014  8.174  8.164       -0.018       8.180           61.175   \n",
      "1291623 -0.012  8.162  8.152       -0.018       8.166           61.075   \n",
      "1291624 -0.014  8.148  8.138       -0.016       8.154           61.050   \n",
      "1291625 -0.014  8.136  2.940       -0.018       8.142           60.975   \n",
      "\n",
      "         Motor_current  COMP  DV_eletric  Towers  MPG  LPS  Pressure_switch  \\\n",
      "1073346         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073347         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073348         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073349         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073350         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "...                ...   ...         ...     ...  ...  ...              ...   \n",
      "1291621         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291622         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291623         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291624         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291625         5.3525   0.0         1.0     1.0  0.0  0.0              1.0   \n",
      "\n",
      "         Oil_level  Caudal_impulses  is_anomaly  \n",
      "1073346        1.0              1.0           0  \n",
      "1073347        1.0              1.0           0  \n",
      "1073348        1.0              1.0           0  \n",
      "1073349        1.0              1.0           0  \n",
      "1073350        1.0              1.0           0  \n",
      "...            ...              ...         ...  \n",
      "1291621        1.0              1.0           0  \n",
      "1291622        1.0              1.0           0  \n",
      "1291623        1.0              1.0           0  \n",
      "1291624        1.0              1.0           0  \n",
      "1291625        1.0              1.0           0  \n",
      "\n",
      "[218280 rows x 16 columns]\n",
      "2020-09-01 03:59:50\n",
      "Train on 752729 samples, validate on 188183 samples\n",
      "Epoch 1/50\n",
      "752729/752729 [==============================] - ETA: 0s - loss: 0.0075"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\anaconda3\\lib\\site-packages\\keras\\src\\engine\\training_v1.py:2335: UserWarning: `Model.state_updates` will be removed in a future version. This property should not be used in TensorFlow 2.0, as `updates` are applied automatically.\n",
      "  updates = self.state_updates\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "752729/752729 [==============================] - 120s 160us/sample - loss: 0.0075 - val_loss: 0.0039\n",
      "Epoch 2/50\n",
      "752729/752729 [==============================] - 117s 155us/sample - loss: 0.0024 - val_loss: 0.0039\n",
      "Epoch 3/50\n",
      "752729/752729 [==============================] - 147s 195us/sample - loss: 0.0016 - val_loss: 0.0037\n",
      "Epoch 4/50\n",
      "752729/752729 [==============================] - 179s 238us/sample - loss: 0.0015 - val_loss: 0.0039\n",
      "Epoch 5/50\n",
      "752729/752729 [==============================] - 163s 216us/sample - loss: 0.0014 - val_loss: 0.0033\n",
      "Epoch 6/50\n",
      "752729/752729 [==============================] - 140s 186us/sample - loss: 0.0012 - val_loss: 0.0033\n",
      "Epoch 7/50\n",
      "752729/752729 [==============================] - 145s 193us/sample - loss: 0.0013 - val_loss: 0.0035\n",
      "Epoch 8/50\n",
      "752729/752729 [==============================] - 147s 196us/sample - loss: 0.0013 - val_loss: 0.0034\n",
      "Epoch 9/50\n",
      "752729/752729 [==============================] - 165s 219us/sample - loss: 0.0016 - val_loss: 0.0036\n",
      "Epoch 10/50\n",
      "752729/752729 [==============================] - 149s 198us/sample - loss: 0.0015 - val_loss: 0.0035\n",
      "Epoch 11/50\n",
      "752729/752729 [==============================] - 127s 169us/sample - loss: 0.0015 - val_loss: 0.0034\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\1723690628.py:128: UserWarning: You are saving your model as an HDF5 file via `model.save()`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')`.\n",
      "  save_model(model, \"vae_digital.h5\")\n",
      "C:\\Users\\vgupt\\AppData\\Local\\anaconda3\\lib\\site-packages\\keras\\src\\engine\\training_v1.py:2359: UserWarning: `Model.state_updates` will be removed in a future version. This property should not be used in TensorFlow 2.0, as `updates` are applied automatically.\n",
      "  updates=self.state_updates,\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of train_pred before reshaping: (940912, 1, 8)\n",
      "Shape of test_pred before reshaping: (218280, 1, 8)\n",
      "Model: \"model_4\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                Output Shape                 Param #   Connected to                  \n",
      "==================================================================================================\n",
      " input_5 (InputLayer)        [(None, 1, 8)]               0         []                            \n",
      "                                                                                                  \n",
      " conv1d_16 (Conv1D)          (None, 1, 32)                800       ['input_5[0][0]']             \n",
      "                                                                                                  \n",
      " conv1d_17 (Conv1D)          (None, 1, 16)                1552      ['conv1d_16[0][0]']           \n",
      "                                                                                                  \n",
      " conv1d_18 (Conv1D)          (None, 1, 6)                 294       ['conv1d_17[0][0]']           \n",
      "                                                                                                  \n",
      " conv1d_19 (Conv1D)          (None, 1, 6)                 294       ['conv1d_17[0][0]']           \n",
      "                                                                                                  \n",
      " lambda_3 (Lambda)           (None, None, 6)              0         ['conv1d_18[0][0]',           \n",
      "                                                                     'conv1d_19[0][0]']           \n",
      "                                                                                                  \n",
      " conv1d_transpose_11 (Conv1  (None, None, 16)             304       ['lambda_3[0][0]']            \n",
      " DTranspose)                                                                                      \n",
      "                                                                                                  \n",
      " conv1d_transpose_12 (Conv1  (None, None, 32)             1568      ['conv1d_transpose_11[0][0]'] \n",
      " DTranspose)                                                                                      \n",
      "                                                                                                  \n",
      " conv1d_20 (Conv1D)          (None, None, 8)              776       ['conv1d_transpose_12[0][0]'] \n",
      "                                                                                                  \n",
      " global_average_pooling1d_3  (None, 8)                    0         ['conv1d_20[0][0]']           \n",
      "  (GlobalAveragePooling1D)                                                                        \n",
      "                                                                                                  \n",
      " reshape_3 (Reshape)         (None, 1, 8)                 0         ['global_average_pooling1d_3[0\n",
      "                                                                    ][0]']                        \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 5588 (21.83 KB)\n",
      "Trainable params: 5588 (21.83 KB)\n",
      "Non-trainable params: 0 (0.00 Byte)\n",
      "__________________________________________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def main():\n",
    "    preprocessor = DataPreprocessor(\"dataset_train_processed.csv\") # change according to where your data is stored\n",
    "    \n",
    "    preprocessor.preprocessing_autoencoder()\n",
    "    autoencoder = AutoencoderModels(preprocessor.digital_train, preprocessor.digital_test) # comment out depending on whether digital/analog data is used\n",
    "    # autoencoder = AutoencoderModels(preprocessor.analog_train, preprocessor.analog_test)\n",
    "    # autoencoder.sae_analog() # comment out according to which method you call from the AutoencoderModels class\n",
    "    autoencoder.vae_digital()\n",
    "    # autoencoder.vae_analog()\n",
    "    # autoencoder.vae_digital()\n",
    "    train_pred, test_pred = autoencoder.autoencoder_predict_digital()\n",
    "    with open(\"vae_digital_train_pred\", \"wb\") as file: # name of file path can change depending on where you want to save it\n",
    "        pickle.dump(train_pred, file)\n",
    "    with open(\"vae_digital_test_pred\", \"wb\") as file: # similarly, this file path can be changed\n",
    "        pickle.dump(test_pred, file)\n",
    "    print(autoencoder.model.summary())\n",
    "    autoencoder.plot_losses()\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "f9fe5c3c",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:53: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.train_data.drop(self.train_data.columns[0], axis=1,\n",
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:55: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data.drop(self.test_data.columns[0], axis=1, inplace=True)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "           TP2    TP3     H1  DV_pressure  Reservoirs  Oil_temperature  \\\n",
      "1073346 -0.012  8.822  8.810       -0.022       8.822           61.825   \n",
      "1073347 -0.012  8.804  8.792       -0.020       8.804           61.625   \n",
      "1073348 -0.012  8.786  8.774       -0.020       8.786           61.575   \n",
      "1073349 -0.012  8.770  8.758       -0.022       8.768           61.450   \n",
      "1073350 -0.012  8.752  8.740       -0.022       8.752           61.350   \n",
      "...        ...    ...    ...          ...         ...              ...   \n",
      "1291621 -0.012  8.188  8.178       -0.018       8.192           61.200   \n",
      "1291622 -0.014  8.174  8.164       -0.018       8.180           61.175   \n",
      "1291623 -0.012  8.162  8.152       -0.018       8.166           61.075   \n",
      "1291624 -0.014  8.148  8.138       -0.016       8.154           61.050   \n",
      "1291625 -0.014  8.136  2.940       -0.018       8.142           60.975   \n",
      "\n",
      "         Motor_current  COMP  DV_eletric  Towers  MPG  LPS  Pressure_switch  \\\n",
      "1073346         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073347         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073348         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073349         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073350         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "...                ...   ...         ...     ...  ...  ...              ...   \n",
      "1291621         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291622         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291623         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291624         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291625         5.3525   0.0         1.0     1.0  0.0  0.0              1.0   \n",
      "\n",
      "         Oil_level  Caudal_impulses  is_anomaly  \n",
      "1073346        1.0              1.0           0  \n",
      "1073347        1.0              1.0           0  \n",
      "1073348        1.0              1.0           0  \n",
      "1073349        1.0              1.0           0  \n",
      "1073350        1.0              1.0           0  \n",
      "...            ...              ...         ...  \n",
      "1291621        1.0              1.0           0  \n",
      "1291622        1.0              1.0           0  \n",
      "1291623        1.0              1.0           0  \n",
      "1291624        1.0              1.0           0  \n",
      "1291625        1.0              1.0           0  \n",
      "\n",
      "[218280 rows x 16 columns]\n",
      "2020-09-01 03:59:50\n",
      "Train on 752729 samples, validate on 188183 samples\n",
      "Epoch 1/50\n",
      "752280/752729 [============================>.] - ETA: 0s - loss: 0.2315"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\anaconda3\\lib\\site-packages\\keras\\src\\engine\\training_v1.py:2335: UserWarning: `Model.state_updates` will be removed in a future version. This property should not be used in TensorFlow 2.0, as `updates` are applied automatically.\n",
      "  updates = self.state_updates\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "752729/752729 [==============================] - 87s 115us/sample - loss: 0.2315 - val_loss: 0.1304\n",
      "Epoch 2/50\n",
      "752729/752729 [==============================] - 85s 113us/sample - loss: 0.1164 - val_loss: 0.0904\n",
      "Epoch 3/50\n",
      "752729/752729 [==============================] - 85s 113us/sample - loss: 0.0891 - val_loss: 0.0785\n",
      "Epoch 4/50\n",
      "752729/752729 [==============================] - 85s 113us/sample - loss: 0.0526 - val_loss: 0.0610\n",
      "Epoch 5/50\n",
      "752729/752729 [==============================] - 92s 122us/sample - loss: 0.0371 - val_loss: 0.0379\n",
      "Epoch 6/50\n",
      "752729/752729 [==============================] - 102s 135us/sample - loss: 0.0209 - val_loss: 0.0113\n",
      "Epoch 7/50\n",
      "752729/752729 [==============================] - 103s 137us/sample - loss: 0.0109 - val_loss: 0.0164\n",
      "Epoch 8/50\n",
      "752729/752729 [==============================] - 111s 147us/sample - loss: 0.0077 - val_loss: 0.0080\n",
      "Epoch 9/50\n",
      "752729/752729 [==============================] - 104s 139us/sample - loss: 0.0068 - val_loss: 0.0118\n",
      "Epoch 10/50\n",
      "752729/752729 [==============================] - 109s 145us/sample - loss: 0.0060 - val_loss: 0.0087\n",
      "Epoch 11/50\n",
      "752729/752729 [==============================] - 94s 125us/sample - loss: 0.0056 - val_loss: 0.0055\n",
      "Epoch 12/50\n",
      "752729/752729 [==============================] - 90s 120us/sample - loss: 0.0053 - val_loss: 0.0084\n",
      "Epoch 13/50\n",
      "752729/752729 [==============================] - 101s 134us/sample - loss: 0.0052 - val_loss: 0.0065\n",
      "Epoch 14/50\n",
      "752729/752729 [==============================] - 108s 143us/sample - loss: 0.0049 - val_loss: 0.0059\n",
      "Epoch 15/50\n",
      "752729/752729 [==============================] - 101s 134us/sample - loss: 0.0042 - val_loss: 0.0057\n",
      "Epoch 16/50\n",
      "752729/752729 [==============================] - 88s 117us/sample - loss: 0.0041 - val_loss: 0.0034\n",
      "Epoch 17/50\n",
      "752729/752729 [==============================] - 100s 133us/sample - loss: 0.0037 - val_loss: 0.0039\n",
      "Epoch 18/50\n",
      "752729/752729 [==============================] - 97s 128us/sample - loss: 0.0036 - val_loss: 0.0035\n",
      "Epoch 19/50\n",
      "752729/752729 [==============================] - 95s 126us/sample - loss: 0.0036 - val_loss: 0.0031\n",
      "Epoch 20/50\n",
      "752729/752729 [==============================] - 93s 124us/sample - loss: 0.0033 - val_loss: 0.0031\n",
      "Epoch 21/50\n",
      "752729/752729 [==============================] - 101s 135us/sample - loss: 0.0031 - val_loss: 0.0028\n",
      "Epoch 22/50\n",
      "752729/752729 [==============================] - 97s 129us/sample - loss: 0.0034 - val_loss: 0.0029\n",
      "Epoch 23/50\n",
      "752729/752729 [==============================] - 98s 130us/sample - loss: 0.0031 - val_loss: 0.0037\n",
      "Epoch 24/50\n",
      "752729/752729 [==============================] - 97s 129us/sample - loss: 0.0029 - val_loss: 0.0024\n",
      "Epoch 25/50\n",
      "752729/752729 [==============================] - 100s 133us/sample - loss: 0.0028 - val_loss: 0.0025\n",
      "Epoch 26/50\n",
      "752729/752729 [==============================] - 95s 126us/sample - loss: 0.0027 - val_loss: 0.0024\n",
      "Epoch 27/50\n",
      "752729/752729 [==============================] - 93s 123us/sample - loss: 0.0028 - val_loss: 0.0024\n",
      "Epoch 28/50\n",
      "752729/752729 [==============================] - 102s 135us/sample - loss: 0.0029 - val_loss: 0.0025\n",
      "Epoch 29/50\n",
      "752729/752729 [==============================] - 95s 126us/sample - loss: 0.0026 - val_loss: 0.0102\n",
      "Epoch 30/50\n",
      "752729/752729 [==============================] - 101s 134us/sample - loss: 0.0027 - val_loss: 0.0026\n",
      "Epoch 31/50\n",
      "752729/752729 [==============================] - 90s 120us/sample - loss: 0.0026 - val_loss: 0.0032\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\1723690628.py:30: UserWarning: You are saving your model as an HDF5 file via `model.save()`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')`.\n",
      "  save_model(model, \"sae_analog.h5\")\n",
      "C:\\Users\\vgupt\\AppData\\Local\\anaconda3\\lib\\site-packages\\keras\\src\\engine\\training_v1.py:2359: UserWarning: `Model.state_updates` will be removed in a future version. This property should not be used in TensorFlow 2.0, as `updates` are applied automatically.\n",
      "  updates=self.state_updates,\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of train_pred before reshaping: (940912, 1, 7)\n",
      "Shape of test_pred before reshaping: (218280, 1, 7)\n",
      "Model: \"model_5\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_6 (InputLayer)        [(None, 1, 7)]            0         \n",
      "                                                                 \n",
      " conv1d_21 (Conv1D)          (None, 1, 32)             704       \n",
      "                                                                 \n",
      " conv1d_22 (Conv1D)          (None, 1, 16)             1552      \n",
      "                                                                 \n",
      " conv1d_23 (Conv1D)          (None, 1, 6)              294       \n",
      "                                                                 \n",
      " conv1d_transpose_13 (Conv1  (None, 1, 16)             304       \n",
      " DTranspose)                                                     \n",
      "                                                                 \n",
      " conv1d_transpose_14 (Conv1  (None, 1, 32)             1568      \n",
      " DTranspose)                                                     \n",
      "                                                                 \n",
      " conv1d_transpose_15 (Conv1  (None, 1, 7)              679       \n",
      " DTranspose)                                                     \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 5101 (19.93 KB)\n",
      "Trainable params: 5101 (19.93 KB)\n",
      "Non-trainable params: 0 (0.00 Byte)\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def main():\n",
    "    preprocessor = DataPreprocessor(\"dataset_train_processed.csv\") # change according to where your data is stored\n",
    "    \n",
    "    preprocessor.preprocessing_autoencoder()\n",
    "    # autoencoder = AutoencoderModels(preprocessor.digital_train, preprocessor.digital_test) # comment out depending on whether digital/analog data is used\n",
    "    autoencoder = AutoencoderModels(preprocessor.analog_train, preprocessor.analog_test)\n",
    "    # autoencoder.sae_analog() # comment out according to which method you call from the AutoencoderModels class\n",
    "    # autoencoder.vae_digital()\n",
    "    autoencoder.sae_analog()\n",
    "    # autoencoder.vae_digital()\n",
    "    train_pred, test_pred = autoencoder.autoencoder_predict_analog()\n",
    "    with open(\"sae_analog_train_pred\", \"wb\") as file: # name of file path can change depending on where you want to save it\n",
    "        pickle.dump(train_pred, file)\n",
    "    with open(\"sae_analog_test_pred\", \"wb\") as file: # similarly, this file path can be changed\n",
    "        pickle.dump(test_pred, file)\n",
    "    print(autoencoder.model.summary())\n",
    "    autoencoder.plot_losses()\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "8dcb2d3a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:53: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.train_data.drop(self.train_data.columns[0], axis=1,\n",
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:55: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data.drop(self.test_data.columns[0], axis=1, inplace=True)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "           TP2    TP3     H1  DV_pressure  Reservoirs  Oil_temperature  \\\n",
      "1073346 -0.012  8.822  8.810       -0.022       8.822           61.825   \n",
      "1073347 -0.012  8.804  8.792       -0.020       8.804           61.625   \n",
      "1073348 -0.012  8.786  8.774       -0.020       8.786           61.575   \n",
      "1073349 -0.012  8.770  8.758       -0.022       8.768           61.450   \n",
      "1073350 -0.012  8.752  8.740       -0.022       8.752           61.350   \n",
      "...        ...    ...    ...          ...         ...              ...   \n",
      "1291621 -0.012  8.188  8.178       -0.018       8.192           61.200   \n",
      "1291622 -0.014  8.174  8.164       -0.018       8.180           61.175   \n",
      "1291623 -0.012  8.162  8.152       -0.018       8.166           61.075   \n",
      "1291624 -0.014  8.148  8.138       -0.016       8.154           61.050   \n",
      "1291625 -0.014  8.136  2.940       -0.018       8.142           60.975   \n",
      "\n",
      "         Motor_current  COMP  DV_eletric  Towers  MPG  LPS  Pressure_switch  \\\n",
      "1073346         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073347         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073348         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073349         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073350         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "...                ...   ...         ...     ...  ...  ...              ...   \n",
      "1291621         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291622         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291623         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291624         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291625         5.3525   0.0         1.0     1.0  0.0  0.0              1.0   \n",
      "\n",
      "         Oil_level  Caudal_impulses  is_anomaly  \n",
      "1073346        1.0              1.0           0  \n",
      "1073347        1.0              1.0           0  \n",
      "1073348        1.0              1.0           0  \n",
      "1073349        1.0              1.0           0  \n",
      "1073350        1.0              1.0           0  \n",
      "...            ...              ...         ...  \n",
      "1291621        1.0              1.0           0  \n",
      "1291622        1.0              1.0           0  \n",
      "1291623        1.0              1.0           0  \n",
      "1291624        1.0              1.0           0  \n",
      "1291625        1.0              1.0           0  \n",
      "\n",
      "[218280 rows x 16 columns]\n",
      "2020-09-01 03:59:50\n",
      "Train on 752729 samples, validate on 188183 samples\n",
      "Epoch 1/50\n",
      "752640/752729 [============================>.] - ETA: 0s - loss: 0.0073"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\anaconda3\\lib\\site-packages\\keras\\src\\engine\\training_v1.py:2335: UserWarning: `Model.state_updates` will be removed in a future version. This property should not be used in TensorFlow 2.0, as `updates` are applied automatically.\n",
      "  updates = self.state_updates\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "752729/752729 [==============================] - 112s 149us/sample - loss: 0.0073 - val_loss: 4.1110e-05\n",
      "Epoch 2/50\n",
      "752729/752729 [==============================] - 97s 128us/sample - loss: 1.5770e-04 - val_loss: 1.9383e-05\n",
      "Epoch 3/50\n",
      "752729/752729 [==============================] - 91s 121us/sample - loss: 9.9785e-05 - val_loss: 1.1846e-05\n",
      "Epoch 4/50\n",
      "752729/752729 [==============================] - 88s 117us/sample - loss: 5.7976e-05 - val_loss: 3.5301e-05\n",
      "Epoch 5/50\n",
      "752729/752729 [==============================] - 92s 122us/sample - loss: 4.0780e-05 - val_loss: 7.1360e-06\n",
      "Epoch 6/50\n",
      "752729/752729 [==============================] - 92s 123us/sample - loss: 4.0514e-05 - val_loss: 2.1781e-05\n",
      "Epoch 7/50\n",
      "752729/752729 [==============================] - 90s 119us/sample - loss: 3.4293e-05 - val_loss: 3.9846e-06\n",
      "Epoch 8/50\n",
      "752729/752729 [==============================] - 87s 115us/sample - loss: 3.5906e-05 - val_loss: 6.6302e-06\n",
      "Epoch 9/50\n",
      "752729/752729 [==============================] - 89s 118us/sample - loss: 3.7692e-05 - val_loss: 8.4152e-06\n",
      "Epoch 10/50\n",
      "752729/752729 [==============================] - 91s 120us/sample - loss: 2.7952e-05 - val_loss: 5.7651e-06\n",
      "Epoch 11/50\n",
      "752729/752729 [==============================] - 92s 123us/sample - loss: 2.6788e-05 - val_loss: 2.8716e-06\n",
      "Epoch 12/50\n",
      "752729/752729 [==============================] - 90s 119us/sample - loss: 3.4326e-05 - val_loss: 9.3715e-06\n",
      "Epoch 13/50\n",
      "752729/752729 [==============================] - 90s 120us/sample - loss: 1.6198e-05 - val_loss: 3.1423e-06\n",
      "Epoch 14/50\n",
      "752729/752729 [==============================] - 89s 118us/sample - loss: 2.8415e-05 - val_loss: 3.0133e-06\n",
      "Epoch 15/50\n",
      "752729/752729 [==============================] - 95s 126us/sample - loss: 1.7659e-05 - val_loss: 2.6444e-06\n",
      "Epoch 16/50\n",
      "752729/752729 [==============================] - 88s 117us/sample - loss: 1.6374e-05 - val_loss: 2.0873e-06\n",
      "Epoch 17/50\n",
      "752729/752729 [==============================] - 90s 119us/sample - loss: 1.9668e-05 - val_loss: 3.8965e-05\n",
      "Epoch 18/50\n",
      "752729/752729 [==============================] - 92s 122us/sample - loss: 2.1815e-05 - val_loss: 5.6308e-06\n",
      "Epoch 19/50\n",
      "752729/752729 [==============================] - 89s 119us/sample - loss: 1.1245e-05 - val_loss: 3.6156e-06\n",
      "Epoch 20/50\n",
      "752729/752729 [==============================] - 87s 115us/sample - loss: 2.2895e-05 - val_loss: 2.3701e-07\n",
      "Epoch 21/50\n",
      "752729/752729 [==============================] - 84s 112us/sample - loss: 1.5707e-05 - val_loss: 4.1445e-07\n",
      "Epoch 22/50\n",
      "752729/752729 [==============================] - 87s 116us/sample - loss: 1.2362e-05 - val_loss: 2.0162e-06\n",
      "Epoch 23/50\n",
      "752729/752729 [==============================] - 89s 119us/sample - loss: 9.7960e-06 - val_loss: 9.6311e-08\n",
      "Epoch 24/50\n",
      "752729/752729 [==============================] - 92s 122us/sample - loss: 1.0366e-05 - val_loss: 8.9204e-07\n",
      "Epoch 25/50\n",
      "752729/752729 [==============================] - 88s 117us/sample - loss: 1.7390e-05 - val_loss: 8.4507e-07\n",
      "Epoch 26/50\n",
      "752729/752729 [==============================] - 87s 115us/sample - loss: 1.0814e-06 - val_loss: 3.9968e-08\n",
      "Epoch 27/50\n",
      "752729/752729 [==============================] - 84s 112us/sample - loss: 8.9325e-06 - val_loss: 1.3746e-07\n",
      "Epoch 28/50\n",
      "752729/752729 [==============================] - 80s 106us/sample - loss: 9.4561e-06 - val_loss: 3.9047e-08\n",
      "Epoch 29/50\n",
      "752729/752729 [==============================] - 86s 114us/sample - loss: 1.2897e-06 - val_loss: 2.5547e-06\n",
      "Epoch 30/50\n",
      "752729/752729 [==============================] - 93s 124us/sample - loss: 1.7019e-06 - val_loss: 5.3479e-09\n",
      "Epoch 31/50\n",
      "752729/752729 [==============================] - 97s 130us/sample - loss: 1.4660e-05 - val_loss: 1.2038e-08\n",
      "Epoch 32/50\n",
      "752729/752729 [==============================] - 91s 121us/sample - loss: 6.3830e-09 - val_loss: 3.4673e-09\n",
      "Epoch 33/50\n",
      "752729/752729 [==============================] - 92s 122us/sample - loss: 2.9807e-05 - val_loss: 1.1932e-08\n",
      "Epoch 34/50\n",
      "752729/752729 [==============================] - 93s 124us/sample - loss: 1.5679e-05 - val_loss: 2.5871e-08\n",
      "Epoch 35/50\n",
      "752729/752729 [==============================] - 89s 118us/sample - loss: 8.3130e-06 - val_loss: 3.9170e-08\n",
      "Epoch 36/50\n",
      "752729/752729 [==============================] - 86s 115us/sample - loss: 3.7944e-07 - val_loss: 8.1139e-10\n",
      "Epoch 37/50\n",
      "752729/752729 [==============================] - 99s 131us/sample - loss: 6.7658e-05 - val_loss: 1.7372e-07\n",
      "Epoch 38/50\n",
      "752729/752729 [==============================] - 90s 120us/sample - loss: 1.3973e-05 - val_loss: 9.5979e-10\n",
      "Epoch 39/50\n",
      "752729/752729 [==============================] - 93s 123us/sample - loss: 5.5222e-06 - val_loss: 1.0788e-09\n",
      "Epoch 40/50\n",
      "752729/752729 [==============================] - 91s 120us/sample - loss: 5.7835e-10 - val_loss: 1.3650e-10\n",
      "Epoch 41/50\n",
      "752729/752729 [==============================] - 95s 126us/sample - loss: 8.2816e-06 - val_loss: 1.5806e-08\n",
      "Epoch 42/50\n",
      "752729/752729 [==============================] - 92s 123us/sample - loss: 2.2939e-05 - val_loss: 1.2082e-09\n",
      "Epoch 43/50\n",
      "752729/752729 [==============================] - 99s 132us/sample - loss: 1.8528e-05 - val_loss: 3.0543e-09\n",
      "Epoch 44/50\n",
      "752729/752729 [==============================] - 97s 130us/sample - loss: 2.1409e-05 - val_loss: 4.6042e-09\n",
      "Epoch 45/50\n",
      "752729/752729 [==============================] - 98s 130us/sample - loss: 1.8035e-09 - val_loss: 2.6617e-10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\1723690628.py:51: UserWarning: You are saving your model as an HDF5 file via `model.save()`. This file format is considered legacy. We recommend using instead the native Keras format, e.g. `model.save('my_model.keras')`.\n",
      "  save_model(model, \"sae_digital.h5\")\n",
      "C:\\Users\\vgupt\\AppData\\Local\\anaconda3\\lib\\site-packages\\keras\\src\\engine\\training_v1.py:2359: UserWarning: `Model.state_updates` will be removed in a future version. This property should not be used in TensorFlow 2.0, as `updates` are applied automatically.\n",
      "  updates=self.state_updates,\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of train_pred before reshaping: (940912, 1, 8)\n",
      "Shape of test_pred before reshaping: (218280, 1, 8)\n",
      "Model: \"model_6\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_7 (InputLayer)        [(None, 1, 8)]            0         \n",
      "                                                                 \n",
      " conv1d_24 (Conv1D)          (None, 1, 32)             800       \n",
      "                                                                 \n",
      " conv1d_25 (Conv1D)          (None, 1, 16)             1552      \n",
      "                                                                 \n",
      " conv1d_26 (Conv1D)          (None, 1, 6)              294       \n",
      "                                                                 \n",
      " conv1d_transpose_16 (Conv1  (None, 1, 16)             304       \n",
      " DTranspose)                                                     \n",
      "                                                                 \n",
      " conv1d_transpose_17 (Conv1  (None, 1, 32)             1568      \n",
      " DTranspose)                                                     \n",
      "                                                                 \n",
      " conv1d_transpose_18 (Conv1  (None, 1, 8)              776       \n",
      " DTranspose)                                                     \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 5294 (20.68 KB)\n",
      "Trainable params: 5294 (20.68 KB)\n",
      "Non-trainable params: 0 (0.00 Byte)\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def main():\n",
    "    preprocessor = DataPreprocessor(\"dataset_train_processed.csv\") # change according to where your data is stored\n",
    "    \n",
    "    preprocessor.preprocessing_autoencoder()\n",
    "    autoencoder = AutoencoderModels(preprocessor.digital_train, preprocessor.digital_test) # comment out depending on whether digital/analog data is used\n",
    "    # autoencoder = AutoencoderModels(preprocessor.analog_train, preprocessor.analog_test)\n",
    "    # autoencoder.sae_analog() # comment out according to which method you call from the AutoencoderModels class\n",
    "    autoencoder.sae_digital()\n",
    "    # autoencoder.vae_analog()\n",
    "    # autoencoder.vae_digital()\n",
    "    train_pred, test_pred = autoencoder.autoencoder_predict_digital()\n",
    "    with open(\"sae_digital_train_pred\", \"wb\") as file: # name of file path can change depending on where you want to save it\n",
    "        pickle.dump(train_pred, file)\n",
    "    with open(\"sae_digital_test_pred\", \"wb\") as file: # similarly, this file path can be changed\n",
    "        pickle.dump(test_pred, file)\n",
    "    print(autoencoder.model.summary())\n",
    "    autoencoder.plot_losses()\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "98e3b228",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "cb7293c5",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pickle\n",
    "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay, classification_report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "29da23ce",
   "metadata": {},
   "outputs": [],
   "source": [
    "np.set_printoptions(suppress=True, precision=10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "25d3c91c",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "1892ba77",
   "metadata": {},
   "outputs": [],
   "source": [
    "class AutoencoderResults:\n",
    "    def __init__(self, train_data, test_data, analog_train, analog_test, digital_train, digital_test):\n",
    "        self.train_data = train_data\n",
    "        self.test_data = test_data\n",
    "        self.analog_train = analog_train\n",
    "        self.analog_test = analog_test\n",
    "        self.digital_train = digital_train\n",
    "        self.digital_test = digital_test\n",
    "        self.train_pred = None\n",
    "        self.test_pred = None\n",
    "        self.threshold = None\n",
    "\n",
    "    def compute_error_sae_digital(self):\n",
    "        with open(\"sae_digital_train_pred\", \"rb\") as file:\n",
    "            self.train_pred = pickle.load(file)\n",
    "        with open(\"sae_digital_test_pred\", \"rb\") as file:\n",
    "            self.test_pred = pickle.load(file)\n",
    "        reconstruction_error = np.mean(np.square(self.train_pred - self.digital_train), axis=1) # can try mse/rmse/mae etc.\n",
    "        q1 = np.percentile(reconstruction_error, 25)\n",
    "        q3 = np.percentile(reconstruction_error, 75)\n",
    "        q9 = np.percentile(reconstruction_error, 99)\n",
    "        self.threshold = q3 + 500000 * (q3 - q1)\n",
    "        # self.threshold = np.mean(reconstruction_error) + 2 * np.std(reconstruction_error)  # can be adjusted to find the optimal threshold\n",
    "        test_error = np.mean(np.square(self.test_pred - self.digital_test), axis=1)\n",
    "        print(q1)\n",
    "        print(q3)\n",
    "        print(max(reconstruction_error))\n",
    "        print(min(test_error))\n",
    "        q1_test = np.percentile(test_error, 25)\n",
    "        q3_test = np.percentile(test_error, 75)\n",
    "        q9_test = np.percentile(test_error, 99)\n",
    "        print(q1_test)\n",
    "        print(q3_test)\n",
    "        print(q9_test)\n",
    "        filtered_error = self.low_pass_filter(test_error,\n",
    "                                         alpha=0.05)  # can be adjusted to find optimal alpha\n",
    "        # predictions = [1 if error > self.threshold else 0 for error in filtered_error]\n",
    "        predictions = [1 if error > 0.1 else 0 for error in filtered_error]\n",
    "        self.test_data[\"predicted_output\"] = predictions\n",
    "        # pred_output = [0] * len(self.test_data)\n",
    "        # for index in self.create_sequence(predictions):\n",
    "        #     pred_output[index] = 1\n",
    "        # self.test_data[\"predicted_output\"] = pred_output\n",
    "        # print(self.test_data[\"is_anomaly\"].value_counts())\n",
    "        # print(self.test_data[\"predicted_output\"].value_counts())\n",
    "        \n",
    "    def compute_error_vae_digital(self):\n",
    "        with open(\"vae_digital_train_pred\", \"rb\") as file:\n",
    "            self.train_pred = pickle.load(file)\n",
    "        with open(\"vae_digital_test_pred\", \"rb\") as file:\n",
    "            self.test_pred = pickle.load(file)\n",
    "        reconstruction_error = np.mean(np.square(self.train_pred - self.digital_train), axis=1) # can try mse/rmse/mae etc.\n",
    "        q1 = np.percentile(reconstruction_error, 25)\n",
    "        q3 = np.percentile(reconstruction_error, 75)\n",
    "        # self.threshold = q3 + 45000 * (q3 - q1)\n",
    "        self.threshold = 0.12\n",
    "        # self.threshold = np.mean(reconstruction_error) + 2 * np.std(reconstruction_error)  # can be adjusted to find the optimal threshold\n",
    "        test_error = np.mean(np.square(self.test_pred - self.digital_test), axis=1)\n",
    "        \n",
    "        filtered_error = self.low_pass_filter(test_error,\n",
    "                                         alpha=0.05)  # can be adjusted to find optimal alpha\n",
    "        predictions = [1 if error > self.threshold else 0 for error in filtered_error]\n",
    "        self.test_data[\"predicted_output\"] = predictions\n",
    "\n",
    "    def compute_error_sae_analog(self):\n",
    "        with open(\"sae_analog_train_pred\", \"rb\") as file:\n",
    "            self.train_pred = pickle.load(file)\n",
    "        with open(\"sae_analog_test_pred\", \"rb\") as file:\n",
    "            self.test_pred = pickle.load(file)\n",
    "        reconstruction_error = np.mean(np.square(self.train_pred - self.analog_train), axis=1) # can try mse/rmse/mae etc.\n",
    "        q1 = np.percentile(reconstruction_error, 25)\n",
    "        q3 = np.percentile(reconstruction_error, 75)\n",
    "        self.threshold = q3 + 2 * (q3 - q1)\n",
    "        # self.threshold = np.mean(reconstruction_error) + 2 * np.std(reconstruction_error)  # can be adjusted to find the optimal threshold\n",
    "        test_error = np.mean(np.square(self.test_pred - self.analog_test), axis=1)\n",
    "        filtered_error = self.low_pass_filter(test_error,\n",
    "                                         alpha=0.1)  # can be adjusted to find optimal alpha\n",
    "        predictions = [1 if error > self.threshold else 0 for error in filtered_error]\n",
    "        self.test_data[\"predicted_output\"] = predictions\n",
    "    \n",
    "    \n",
    "    def compute_error_vae_analog(self):\n",
    "        with open(\"vae_analog_train_pred\", \"rb\") as file:\n",
    "            self.train_pred = pickle.load(file)\n",
    "        with open(\"vae_analog_test_pred\", \"rb\") as file:\n",
    "            self.test_pred = pickle.load(file)\n",
    "        reconstruction_error = np.mean(np.square(self.train_pred - self.analog_train), axis=1) # can try mse/rmse/mae etc.\n",
    "        q1 = np.percentile(reconstruction_error, 25)\n",
    "        q3 = np.percentile(reconstruction_error, 75)\n",
    "        self.threshold = q3 + 2 * (q3 - q1)\n",
    "        # self.threshold = np.mean(reconstruction_error) + 2 * np.std(reconstruction_error)  # can be adjusted to find the optimal threshold\n",
    "        test_error = np.mean(np.square(self.test_pred - self.analog_test), axis=1)\n",
    "        filtered_error = self.low_pass_filter(test_error,\n",
    "                                         alpha=0.1)  # can be adjusted to find optimal alpha\n",
    "        predictions = [1 if error > self.threshold else 0 for error in filtered_error]\n",
    "        self.test_data[\"predicted_output\"] = predictions\n",
    "    \n",
    "    \n",
    "    def low_pass_filter(self, error, alpha): # test_error is passed through low_pass_filter to improve evaluation metrics\n",
    "        error = error.reset_index(drop = True)\n",
    "        output = [error[0]]  # initialize the output with the first input value\n",
    "        for i in range(1, len(error)):\n",
    "            output.append(output[i - 1] + alpha * (error[i] - output[i - 1]))\n",
    "        return output\n",
    "\n",
    "    def create_sequence(self, anomalies):  # check if anomaly data points appear consecutively\n",
    "    # if a point appears by itself, it has a high chance of simply being random noise rather than a part of any failure\n",
    "    # returns index of anomalies\n",
    "        tmp1 = []\n",
    "        tmp2 = []\n",
    "        lst = []\n",
    "        for i in range(len(anomalies)):\n",
    "            if anomalies[i] == 1:\n",
    "                tmp1.append(i)\n",
    "        for j in range(len(tmp1) - 1):\n",
    "            if tmp1[j] + 1 == tmp1[j + 1]:  # check if the next index is the same as the previous index + 1 (eg. 501 == 500 + 1)\n",
    "                tmp2.append(tmp1[j])\n",
    "            if tmp1[j + 1] != tmp1[j] + 1:\n",
    "                if len(tmp2) >= 2:  # sequences have to be x seconds or more - can be adjusted to affect precision/recall\n",
    "                    tmp2.append(tmp1[j])\n",
    "                    lst.extend(tmp2)\n",
    "                tmp2 = []\n",
    "        return lst\n",
    "\n",
    "    def plot_cm_SAE_Digital(self):\n",
    "        cm = confusion_matrix(self.test_data[\"is_anomaly\"], self.test_data[\"predicted_output\"])\n",
    "        tp = cm[1, 1]\n",
    "        tn = cm[0, 0]\n",
    "        fp = cm[0, 1]\n",
    "        fn = cm[1, 0]\n",
    "        print(classification_report(self.test_data[\"is_anomaly\"], self.test_data[\"predicted_output\"]))\n",
    "        precision = tp / (tp + fp)\n",
    "        recall = tp / (tp + fn)\n",
    "        f1_score = 2 * (precision * recall) / (precision + recall)\n",
    "        disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=[\"Normal\", \"Anomaly\"])\n",
    "        disp.plot(values_format='d', cmap=\"Reds\")\n",
    "        plt.title(\"Sparse Autoencoder - Digital (alpha=0.05)\\nPrecision={:.2f}, Recall={:.2f}, F1 Score={:.2f}\".format(precision, recall, f1_score))\n",
    "        plt.show()\n",
    "        \n",
    "        \n",
    "    def plot_cm_VAE_Digital(self):\n",
    "        cm = confusion_matrix(self.test_data[\"is_anomaly\"], self.test_data[\"predicted_output\"])\n",
    "        tp = cm[1, 1]\n",
    "        tn = cm[0, 0]\n",
    "        fp = cm[0, 1]\n",
    "        fn = cm[1, 0]\n",
    "        print(classification_report(self.test_data[\"is_anomaly\"], self.test_data[\"predicted_output\"]))\n",
    "        precision = tp / (tp + fp)\n",
    "        recall = tp / (tp + fn)\n",
    "        f1_score = 2 * (precision * recall) / (precision + recall)\n",
    "        disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=[\"Normal\", \"Anomaly\"])\n",
    "        disp.plot(values_format='d', cmap=\"Reds\")\n",
    "        plt.title(\"Variational Autoencoder - Digital (alpha=0.05)\\nPrecision={:.2f}, Recall={:.2f}, F1 Score={:.2f}\".format(precision, recall, f1_score))\n",
    "        plt.show()\n",
    "        \n",
    "        \n",
    "    def plot_cm_SAE_Analog(self):\n",
    "        cm = confusion_matrix(self.test_data[\"is_anomaly\"], self.test_data[\"predicted_output\"])\n",
    "        tp = cm[1, 1]\n",
    "        tn = cm[0, 0]\n",
    "        fp = cm[0, 1]\n",
    "        fn = cm[1, 0]\n",
    "        print(classification_report(self.test_data[\"is_anomaly\"], self.test_data[\"predicted_output\"]))\n",
    "        precision = tp / (tp + fp)\n",
    "        recall = tp / (tp + fn)\n",
    "        f1_score = 2 * (precision * recall) / (precision + recall)\n",
    "        disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=[\"Normal\", \"Anomaly\"])\n",
    "        disp.plot(values_format='d', cmap=\"Reds\")\n",
    "        plt.title(\"Sparse Autoencoder - Analog (alpha=0.05)\\nPrecision={:.2f}, Recall={:.2f}, F1 Score={:.2f}\".format(precision, recall, f1_score))\n",
    "        plt.show()\n",
    "        \n",
    "        \n",
    "    def plot_cm_VAE_Analog(self):\n",
    "        cm = confusion_matrix(self.test_data[\"is_anomaly\"], self.test_data[\"predicted_output\"])\n",
    "        tp = cm[1, 1]\n",
    "        tn = cm[0, 0]\n",
    "        fp = cm[0, 1]\n",
    "        fn = cm[1, 0]\n",
    "        print(classification_report(self.test_data[\"is_anomaly\"], self.test_data[\"predicted_output\"]))\n",
    "        precision = tp / (tp + fp)\n",
    "        recall = tp / (tp + fn)\n",
    "        f1_score = 2 * (precision * recall) / (precision + recall)\n",
    "        disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=[\"Normal\", \"Anomaly\"])\n",
    "        disp.plot(values_format='d', cmap=\"Reds\")\n",
    "        plt.title(\"Variational Autoencoder - Analog (alpha=0.05)\\nPrecision={:.2f}, Recall={:.2f}, F1 Score={:.2f}\".format(precision, recall, f1_score))\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "a777a82c",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:53: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.train_data.drop(self.train_data.columns[0], axis=1,\n",
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:55: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data.drop(self.test_data.columns[0], axis=1, inplace=True)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "           TP2    TP3     H1  DV_pressure  Reservoirs  Oil_temperature  \\\n",
      "1073346 -0.012  8.822  8.810       -0.022       8.822           61.825   \n",
      "1073347 -0.012  8.804  8.792       -0.020       8.804           61.625   \n",
      "1073348 -0.012  8.786  8.774       -0.020       8.786           61.575   \n",
      "1073349 -0.012  8.770  8.758       -0.022       8.768           61.450   \n",
      "1073350 -0.012  8.752  8.740       -0.022       8.752           61.350   \n",
      "...        ...    ...    ...          ...         ...              ...   \n",
      "1291621 -0.012  8.188  8.178       -0.018       8.192           61.200   \n",
      "1291622 -0.014  8.174  8.164       -0.018       8.180           61.175   \n",
      "1291623 -0.012  8.162  8.152       -0.018       8.166           61.075   \n",
      "1291624 -0.014  8.148  8.138       -0.016       8.154           61.050   \n",
      "1291625 -0.014  8.136  2.940       -0.018       8.142           60.975   \n",
      "\n",
      "         Motor_current  COMP  DV_eletric  Towers  MPG  LPS  Pressure_switch  \\\n",
      "1073346         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073347         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073348         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073349         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073350         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "...                ...   ...         ...     ...  ...  ...              ...   \n",
      "1291621         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291622         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291623         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291624         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291625         5.3525   0.0         1.0     1.0  0.0  0.0              1.0   \n",
      "\n",
      "         Oil_level  Caudal_impulses  is_anomaly  \n",
      "1073346        1.0              1.0           0  \n",
      "1073347        1.0              1.0           0  \n",
      "1073348        1.0              1.0           0  \n",
      "1073349        1.0              1.0           0  \n",
      "1073350        1.0              1.0           0  \n",
      "...            ...              ...         ...  \n",
      "1291621        1.0              1.0           0  \n",
      "1291622        1.0              1.0           0  \n",
      "1291623        1.0              1.0           0  \n",
      "1291624        1.0              1.0           0  \n",
      "1291625        1.0              1.0           0  \n",
      "\n",
      "[218280 rows x 16 columns]\n",
      "2020-09-01 03:59:50\n",
      "9.013412946289928e-47\n",
      "9.013412946289928e-47\n",
      "1.4075359957418707e-09\n",
      "0.0\n",
      "9.013412946289928e-47\n",
      "9.013412946289928e-47\n",
      "0.125\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\2754281283.py:39: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data[\"predicted_output\"] = predictions\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      1.00      1.00    201518\n",
      "           1       1.00      0.99      1.00     16762\n",
      "\n",
      "    accuracy                           1.00    218280\n",
      "   macro avg       1.00      1.00      1.00    218280\n",
      "weighted avg       1.00      1.00      1.00    218280\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# sae_digital\n",
    "\n",
    "def main():\n",
    "    preprocessor = DataPreprocessor(\"dataset_train_processed.csv\")\n",
    "    preprocessor.preprocessing_autoencoder()\n",
    "    autoencoder_results = AutoencoderResults(preprocessor.train_data, preprocessor.test_data, preprocessor.analog_train,\n",
    "                                             preprocessor.analog_test, preprocessor.digital_train, preprocessor.digital_test)\n",
    "    autoencoder_results.compute_error_sae_digital()\n",
    "    autoencoder_results.plot_cm_SAE_Digital()\n",
    "\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "1c181355",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:53: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.train_data.drop(self.train_data.columns[0], axis=1,\n",
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:55: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data.drop(self.test_data.columns[0], axis=1, inplace=True)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "           TP2    TP3     H1  DV_pressure  Reservoirs  Oil_temperature  \\\n",
      "1073346 -0.012  8.822  8.810       -0.022       8.822           61.825   \n",
      "1073347 -0.012  8.804  8.792       -0.020       8.804           61.625   \n",
      "1073348 -0.012  8.786  8.774       -0.020       8.786           61.575   \n",
      "1073349 -0.012  8.770  8.758       -0.022       8.768           61.450   \n",
      "1073350 -0.012  8.752  8.740       -0.022       8.752           61.350   \n",
      "...        ...    ...    ...          ...         ...              ...   \n",
      "1291621 -0.012  8.188  8.178       -0.018       8.192           61.200   \n",
      "1291622 -0.014  8.174  8.164       -0.018       8.180           61.175   \n",
      "1291623 -0.012  8.162  8.152       -0.018       8.166           61.075   \n",
      "1291624 -0.014  8.148  8.138       -0.016       8.154           61.050   \n",
      "1291625 -0.014  8.136  2.940       -0.018       8.142           60.975   \n",
      "\n",
      "         Motor_current  COMP  DV_eletric  Towers  MPG  LPS  Pressure_switch  \\\n",
      "1073346         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073347         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073348         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073349         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073350         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "...                ...   ...         ...     ...  ...  ...              ...   \n",
      "1291621         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291622         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291623         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291624         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291625         5.3525   0.0         1.0     1.0  0.0  0.0              1.0   \n",
      "\n",
      "         Oil_level  Caudal_impulses  is_anomaly  \n",
      "1073346        1.0              1.0           0  \n",
      "1073347        1.0              1.0           0  \n",
      "1073348        1.0              1.0           0  \n",
      "1073349        1.0              1.0           0  \n",
      "1073350        1.0              1.0           0  \n",
      "...            ...              ...         ...  \n",
      "1291621        1.0              1.0           0  \n",
      "1291622        1.0              1.0           0  \n",
      "1291623        1.0              1.0           0  \n",
      "1291624        1.0              1.0           0  \n",
      "1291625        1.0              1.0           0  \n",
      "\n",
      "[218280 rows x 16 columns]\n",
      "2020-09-01 03:59:50\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\2754281283.py:63: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data[\"predicted_output\"] = predictions\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      0.97      0.99    201518\n",
      "           1       0.74      1.00      0.85     16762\n",
      "\n",
      "    accuracy                           0.97    218280\n",
      "   macro avg       0.87      0.98      0.92    218280\n",
      "weighted avg       0.98      0.97      0.97    218280\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# vae_digital\n",
    "\n",
    "def main():\n",
    "    preprocessor = DataPreprocessor(\"dataset_train_processed.csv\")\n",
    "    preprocessor.preprocessing_autoencoder()\n",
    "    autoencoder_results = AutoencoderResults(preprocessor.train_data, preprocessor.test_data, preprocessor.analog_train,\n",
    "                                             preprocessor.analog_test, preprocessor.digital_train, preprocessor.digital_test)\n",
    "    autoencoder_results.compute_error_vae_digital()\n",
    "    autoencoder_results.plot_cm_VAE_Digital()\n",
    "\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "6764b98a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:53: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.train_data.drop(self.train_data.columns[0], axis=1,\n",
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:55: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data.drop(self.test_data.columns[0], axis=1, inplace=True)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "           TP2    TP3     H1  DV_pressure  Reservoirs  Oil_temperature  \\\n",
      "1073346 -0.012  8.822  8.810       -0.022       8.822           61.825   \n",
      "1073347 -0.012  8.804  8.792       -0.020       8.804           61.625   \n",
      "1073348 -0.012  8.786  8.774       -0.020       8.786           61.575   \n",
      "1073349 -0.012  8.770  8.758       -0.022       8.768           61.450   \n",
      "1073350 -0.012  8.752  8.740       -0.022       8.752           61.350   \n",
      "...        ...    ...    ...          ...         ...              ...   \n",
      "1291621 -0.012  8.188  8.178       -0.018       8.192           61.200   \n",
      "1291622 -0.014  8.174  8.164       -0.018       8.180           61.175   \n",
      "1291623 -0.012  8.162  8.152       -0.018       8.166           61.075   \n",
      "1291624 -0.014  8.148  8.138       -0.016       8.154           61.050   \n",
      "1291625 -0.014  8.136  2.940       -0.018       8.142           60.975   \n",
      "\n",
      "         Motor_current  COMP  DV_eletric  Towers  MPG  LPS  Pressure_switch  \\\n",
      "1073346         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073347         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073348         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073349         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073350         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "...                ...   ...         ...     ...  ...  ...              ...   \n",
      "1291621         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291622         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291623         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291624         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291625         5.3525   0.0         1.0     1.0  0.0  0.0              1.0   \n",
      "\n",
      "         Oil_level  Caudal_impulses  is_anomaly  \n",
      "1073346        1.0              1.0           0  \n",
      "1073347        1.0              1.0           0  \n",
      "1073348        1.0              1.0           0  \n",
      "1073349        1.0              1.0           0  \n",
      "1073350        1.0              1.0           0  \n",
      "...            ...              ...         ...  \n",
      "1291621        1.0              1.0           0  \n",
      "1291622        1.0              1.0           0  \n",
      "1291623        1.0              1.0           0  \n",
      "1291624        1.0              1.0           0  \n",
      "1291625        1.0              1.0           0  \n",
      "\n",
      "[218280 rows x 16 columns]\n",
      "2020-09-01 03:59:50\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\2754281283.py:79: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data[\"predicted_output\"] = predictions\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.77      0.85    201518\n",
      "           1       0.18      0.60      0.27     16762\n",
      "\n",
      "    accuracy                           0.75    218280\n",
      "   macro avg       0.57      0.68      0.56    218280\n",
      "weighted avg       0.90      0.75      0.81    218280\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# sae_analog\n",
    "\n",
    "def main():\n",
    "    preprocessor = DataPreprocessor(\"dataset_train_processed.csv\")\n",
    "    preprocessor.preprocessing_autoencoder()\n",
    "    autoencoder_results = AutoencoderResults(preprocessor.train_data, preprocessor.test_data, preprocessor.analog_train,\n",
    "                                             preprocessor.analog_test, preprocessor.digital_train, preprocessor.digital_test)\n",
    "    autoencoder_results.compute_error_sae_analog()\n",
    "    autoencoder_results.plot_cm_SAE_Analog()\n",
    "\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "37f090de",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:53: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.train_data.drop(self.train_data.columns[0], axis=1,\n",
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\259275283.py:55: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data.drop(self.test_data.columns[0], axis=1, inplace=True)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "           TP2    TP3     H1  DV_pressure  Reservoirs  Oil_temperature  \\\n",
      "1073346 -0.012  8.822  8.810       -0.022       8.822           61.825   \n",
      "1073347 -0.012  8.804  8.792       -0.020       8.804           61.625   \n",
      "1073348 -0.012  8.786  8.774       -0.020       8.786           61.575   \n",
      "1073349 -0.012  8.770  8.758       -0.022       8.768           61.450   \n",
      "1073350 -0.012  8.752  8.740       -0.022       8.752           61.350   \n",
      "...        ...    ...    ...          ...         ...              ...   \n",
      "1291621 -0.012  8.188  8.178       -0.018       8.192           61.200   \n",
      "1291622 -0.014  8.174  8.164       -0.018       8.180           61.175   \n",
      "1291623 -0.012  8.162  8.152       -0.018       8.166           61.075   \n",
      "1291624 -0.014  8.148  8.138       -0.016       8.154           61.050   \n",
      "1291625 -0.014  8.136  2.940       -0.018       8.142           60.975   \n",
      "\n",
      "         Motor_current  COMP  DV_eletric  Towers  MPG  LPS  Pressure_switch  \\\n",
      "1073346         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073347         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073348         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073349         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1073350         0.0425   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "...                ...   ...         ...     ...  ...  ...              ...   \n",
      "1291621         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291622         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291623         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291624         0.0375   1.0         0.0     1.0  1.0  0.0              1.0   \n",
      "1291625         5.3525   0.0         1.0     1.0  0.0  0.0              1.0   \n",
      "\n",
      "         Oil_level  Caudal_impulses  is_anomaly  \n",
      "1073346        1.0              1.0           0  \n",
      "1073347        1.0              1.0           0  \n",
      "1073348        1.0              1.0           0  \n",
      "1073349        1.0              1.0           0  \n",
      "1073350        1.0              1.0           0  \n",
      "...            ...              ...         ...  \n",
      "1291621        1.0              1.0           0  \n",
      "1291622        1.0              1.0           0  \n",
      "1291623        1.0              1.0           0  \n",
      "1291624        1.0              1.0           0  \n",
      "1291625        1.0              1.0           0  \n",
      "\n",
      "[218280 rows x 16 columns]\n",
      "2020-09-01 03:59:50\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\vgupt\\AppData\\Local\\Temp\\ipykernel_9336\\2754281283.py:96: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self.test_data[\"predicted_output\"] = predictions\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      0.90      0.93    201518\n",
      "           1       0.26      0.41      0.32     16762\n",
      "\n",
      "    accuracy                           0.87    218280\n",
      "   macro avg       0.61      0.66      0.62    218280\n",
      "weighted avg       0.90      0.87      0.88    218280\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# vae_analog\n",
    "\n",
    "def main():\n",
    "    preprocessor = DataPreprocessor(\"dataset_train_processed.csv\")\n",
    "    preprocessor.preprocessing_autoencoder()\n",
    "    autoencoder_results = AutoencoderResults(preprocessor.train_data, preprocessor.test_data, preprocessor.analog_train,\n",
    "                                             preprocessor.analog_test, preprocessor.digital_train, preprocessor.digital_test)\n",
    "    autoencoder_results.compute_error_vae_analog()\n",
    "    autoencoder_results.plot_cm_VAE_Analog()\n",
    "\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    main()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "236bcefd",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5ca93db7",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6aeea231",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}