Remaining_Life_Estimation_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "pTttwUGpitR2"
   },
   "outputs": [],
   "source": [
    "#!pip install sqlalchemy\n",
    "import json\n",
    "import os\n",
    "import itertools\n",
    "from datetime import datetime\n",
    "import math\n",
    "from sqlalchemy import create_engine"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "r-5cLxACitSA"
   },
   "outputs": [],
   "source": [
    "from numpy import array\n",
    "from numpy import argmax\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from keras.models import Sequential, load_model\n",
    "from keras.layers import Dense, LSTM, Dropout\n",
    "from keras.backend import clear_session\n",
    "from keras import backend as K"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "nqHCYFJzitSC"
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "#!pip install pickle5\n",
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "KTy-7jupitSD"
   },
   "outputs": [],
   "source": [
    "#!pip install colorcet\n",
    "import pandas as pd\n",
    "from pandas.api.types import CategoricalDtype \n",
    "import numpy as np\n",
    "import colorcet as cc\n",
    "from numpy import linspace\n",
    "from scipy.stats.kde import gaussian_kde"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 51,
     "status": "ok",
     "timestamp": 1632917677522,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "qadiHyLoitSE",
    "outputId": "4a732524-5eb1-4d9d-9536-9ac83b177908"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
     ]
    }
   ],
   "source": [
    "#leggo i dataframe dal disco\n",
    "from google.colab import drive\n",
    "drive.mount('/content/drive')\n",
    "DATA_PATH = 'drive/MyDrive/Colab Notebooks/PHP4/Remaining_Life_Estimation'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "ffgLZVGKjPUo"
   },
   "outputs": [],
   "source": [
    "train_df = pd.read_pickle(DATA_PATH + '/train_df.pkl')\n",
    "test_df = pd.read_pickle(DATA_PATH + '/test_df.pkl')\n",
    "train_df_y = pd.read_pickle(DATA_PATH + '/train_df_y.pkl')\n",
    "test_df_y = pd.read_pickle(DATA_PATH + '/test_df_y.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "JX1M-JYGkk-H"
   },
   "outputs": [],
   "source": [
    "#!ls 'drive/MyDrive/Colab Notebooks'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 39,
     "status": "ok",
     "timestamp": 1632917677525,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "JXLah14eitSF",
    "outputId": "1847c794-9e15-4036-f20e-bece67250f76"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "31"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_steps = 50\n",
    "batch_size = 1\n",
    "num_features_x = 24\n",
    "num_features_y = 3\n",
    "\n",
    "n_rows = train_df.shape[0] % (num_steps * batch_size)\n",
    "n_rows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 33,
     "status": "ok",
     "timestamp": 1632917677526,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "mJSMaXz9L77u",
    "outputId": "b595346c-962c-44ba-8ce5-6f2f2b06cf07"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "20600"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.drop(train_df.tail(n_rows).index, inplace = True)\n",
    "train_df_y.drop(train_df_y.tail(n_rows).index, inplace = True)\n",
    "n_rows = test_df.shape[0] % (num_steps * batch_size)\n",
    "test_df.drop(test_df.tail(n_rows).index, inplace = True)\n",
    "test_df_y.drop(test_df_y.tail(n_rows).index, inplace = True)\n",
    "train_df.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "IREelfv-itSH"
   },
   "outputs": [],
   "source": [
    "#train_df.isnull().values.any()\n",
    "#pd.set_option(\"display.max_columns\", None)\n",
    "#train_df[train_df.isnull().any(axis=1)]\n",
    "train_df = train_df.fillna(0)\n",
    "test_df = test_df.fillna(0)\n",
    "#pd.reset_option(\"display.max_columns\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "TQVyWrFZitSI"
   },
   "outputs": [],
   "source": [
    "x_train = train_df.drop(['id', 'cycle'],axis=1).to_numpy()\n",
    "x_test = test_df.drop(['id', 'cycle'],axis=1).to_numpy()\n",
    "y_train = train_df_y.drop(['id', 'cycle'],axis=1).to_numpy()\n",
    "y_test = test_df_y.drop(['id', 'cycle'],axis=1).to_numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 29,
     "status": "ok",
     "timestamp": 1632917677529,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "v2msrJW6Mi8R",
    "outputId": "4256dcdd-cf2f-424e-ed85-353775b4a41c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.31597968, -1.3729532 ,  0.        , ...,  0.        ,\n",
       "         1.34849274,  1.19442705],\n",
       "       [ 0.87272193, -1.03172035,  0.        , ...,  0.        ,\n",
       "         1.01652793,  1.23692196],\n",
       "       [-1.96187421,  1.01567674,  0.        , ...,  0.        ,\n",
       "         0.73989059,  0.50342281],\n",
       "       ...,\n",
       "       [ 0.82700264, -1.3729532 ,  0.        , ...,  0.        ,\n",
       "        -1.14124336, -0.57465465],\n",
       "       [-1.23036553, -0.69048751,  0.        , ...,  0.        ,\n",
       "        -1.14124336, -0.25132379],\n",
       "       [-0.86461119, -1.03172035,  0.        , ...,  0.        ,\n",
       "        -0.58796867, -0.67996813]])"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "nbgtTWYCitSJ"
   },
   "outputs": [],
   "source": [
    "x_train_shaped = np.reshape(x_train, newshape=(-1, num_steps, num_features_x))\n",
    "x_test_shaped = np.reshape(x_test, newshape=(-1, num_steps, num_features_x))\n",
    "y_train_shaped = np.reshape(y_train, newshape=(-1, num_steps, num_features_y))\n",
    "y_test_shaped = np.reshape(y_test, newshape=(-1, num_steps, num_features_y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 21,
     "status": "ok",
     "timestamp": 1632917677721,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "p4036MkkitSJ",
    "outputId": "e6146e8a-4513-4c8b-ee78-7366e08287b4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(412, 50, 24)\n",
      "(412, 50, 3)\n",
      "(261, 50, 24)\n",
      "(261, 50, 3)\n"
     ]
    }
   ],
   "source": [
    "print(x_train_shaped.shape)\n",
    "print(y_train_shaped.shape)\n",
    "print(x_test_shaped.shape)\n",
    "print(y_test_shaped.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "xKq-9NjhitSK"
   },
   "outputs": [],
   "source": [
    "x_train_final = np.asarray(x_train_shaped).astype('float32')\n",
    "y_train_final = np.asarray(y_train_shaped).astype('float32')\n",
    "x_test_final = np.asarray(x_test_shaped).astype('float32')\n",
    "y_test_final = np.asarray(y_test_shaped).astype('float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "17WGa5U4zq2x"
   },
   "outputs": [],
   "source": [
    "def weighted_categorical_crossentropy(weights):\n",
    "    \"\"\"\n",
    "    A weighted version of keras.objectives.categorical_crossentropy\n",
    "    \n",
    "    Variables:\n",
    "        weights: numpy array of shape (C,) where C is the number of classes\n",
    "    \n",
    "    Usage:\n",
    "        weights = np.array([0.5,2,10]) # Class one at 0.5, class 2 twice the normal weights, class 3 10x.\n",
    "        loss = weighted_categorical_crossentropy(weights)\n",
    "        model.compile(loss=loss,optimizer='adam')\n",
    "    \"\"\"\n",
    "    \n",
    "    weights = K.variable(weights)\n",
    "        \n",
    "    def loss(y_true, y_pred):\n",
    "        # scale predictions so that the class probas of each sample sum to 1\n",
    "        y_pred /= K.sum(y_pred, axis=-1, keepdims=True)\n",
    "        # clip to prevent NaN's and Inf's\n",
    "        y_pred = K.clip(y_pred, K.epsilon(), 1 - K.epsilon())\n",
    "        # calc\n",
    "        loss = y_true * K.log(y_pred) * weights\n",
    "        loss = -K.sum(loss, -1)\n",
    "        return loss\n",
    "    \n",
    "    return loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "BqdgouS1z1o2"
   },
   "outputs": [],
   "source": [
    "weights = np.array([0.5,5,10]) # Class one at 0.5, class 2 twice the normal weights, class 3 10x.\n",
    "loss = weighted_categorical_crossentropy(weights)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 1118,
     "status": "ok",
     "timestamp": 1632917679395,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "OwcgaRklitSL",
    "outputId": "5f52e5e1-f6ed-4b61-82ae-6fd40b6f9f60"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "lstm (LSTM)                  (1, 50, 50)               15000     \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (1, 50, 50)               0         \n",
      "_________________________________________________________________\n",
      "lstm_1 (LSTM)                (1, 50, 60)               26640     \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (1, 50, 60)               0         \n",
      "_________________________________________________________________\n",
      "lstm_2 (LSTM)                (1, 50, 60)               29040     \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (1, 50, 60)               0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (1, 50, 3)                183       \n",
      "=================================================================\n",
      "Total params: 70,863\n",
      "Trainable params: 70,863\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "#model\n",
    "model = Sequential()\n",
    "#input\n",
    "model.add(LSTM(units=50, return_sequences=True, activation='tanh', batch_size=batch_size, stateful=True, input_shape = (x_train_final.shape[1], x_train_final.shape[2])))\n",
    "model.add(Dropout(0.2))\n",
    "#hidden layer 1\n",
    "model.add(LSTM(units=60, return_sequences=True, activation='tanh', stateful=True))\n",
    "model.add(Dropout(0.2))\n",
    "#hidden layer 2\n",
    "model.add(LSTM(units=60, return_sequences=True, activation='tanh', stateful=True))\n",
    "model.add(Dropout(0.2))\n",
    "#output\n",
    "model.add(Dense(units=3,activation='softmax'))\n",
    "#model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "model.compile(loss=loss, optimizer='adam', metrics=['accuracy'])\n",
    "#summary\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 563337,
     "status": "ok",
     "timestamp": 1632920441666,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "5ytjiixPitSN",
    "outputId": "5f892989-bc72-42d2-feca-9327c516f000"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "412/412 [==============================] - 12s 27ms/step - loss: 0.1907 - accuracy: 0.9650\n",
      "Epoch 2/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0733 - accuracy: 0.9812\n",
      "Epoch 3/50\n",
      "412/412 [==============================] - 11s 28ms/step - loss: 0.0442 - accuracy: 0.9897\n",
      "Epoch 4/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0440 - accuracy: 0.9901\n",
      "Epoch 5/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0432 - accuracy: 0.9880\n",
      "Epoch 6/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0383 - accuracy: 0.9907\n",
      "Epoch 7/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0425 - accuracy: 0.9890\n",
      "Epoch 8/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0387 - accuracy: 0.9930\n",
      "Epoch 9/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0723 - accuracy: 0.9792\n",
      "Epoch 10/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0663 - accuracy: 0.9841\n",
      "Epoch 11/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0416 - accuracy: 0.9883\n",
      "Epoch 12/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0308 - accuracy: 0.9929\n",
      "Epoch 13/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0784 - accuracy: 0.9908\n",
      "Epoch 14/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0602 - accuracy: 0.9899\n",
      "Epoch 15/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0404 - accuracy: 0.9892\n",
      "Epoch 16/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.1125 - accuracy: 0.9826\n",
      "Epoch 17/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0747 - accuracy: 0.9867\n",
      "Epoch 18/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0669 - accuracy: 0.9823\n",
      "Epoch 19/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0304 - accuracy: 0.9917\n",
      "Epoch 20/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0255 - accuracy: 0.9925\n",
      "Epoch 21/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0263 - accuracy: 0.9931\n",
      "Epoch 22/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0739 - accuracy: 0.9818\n",
      "Epoch 23/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0586 - accuracy: 0.9883\n",
      "Epoch 24/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0474 - accuracy: 0.9900\n",
      "Epoch 25/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0262 - accuracy: 0.9926\n",
      "Epoch 26/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0325 - accuracy: 0.9923\n",
      "Epoch 27/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0929 - accuracy: 0.9867\n",
      "Epoch 28/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0424 - accuracy: 0.9909\n",
      "Epoch 29/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0732 - accuracy: 0.9875\n",
      "Epoch 30/50\n",
      "412/412 [==============================] - 11s 28ms/step - loss: 0.0381 - accuracy: 0.9877\n",
      "Epoch 31/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0321 - accuracy: 0.9908\n",
      "Epoch 32/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0282 - accuracy: 0.9935\n",
      "Epoch 33/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0366 - accuracy: 0.9910\n",
      "Epoch 34/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0248 - accuracy: 0.9918\n",
      "Epoch 35/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0293 - accuracy: 0.9937\n",
      "Epoch 36/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0339 - accuracy: 0.9924\n",
      "Epoch 37/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.1558 - accuracy: 0.9834\n",
      "Epoch 38/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0471 - accuracy: 0.9903\n",
      "Epoch 39/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0327 - accuracy: 0.9918\n",
      "Epoch 40/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0650 - accuracy: 0.9901\n",
      "Epoch 41/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0410 - accuracy: 0.9908\n",
      "Epoch 42/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0347 - accuracy: 0.9913\n",
      "Epoch 43/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0294 - accuracy: 0.9935\n",
      "Epoch 44/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0193 - accuracy: 0.9957\n",
      "Epoch 45/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0234 - accuracy: 0.9954\n",
      "Epoch 46/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0221 - accuracy: 0.9952\n",
      "Epoch 47/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0277 - accuracy: 0.9934\n",
      "Epoch 48/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0966 - accuracy: 0.9787\n",
      "Epoch 49/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0671 - accuracy: 0.9858\n",
      "Epoch 50/50\n",
      "412/412 [==============================] - 11s 27ms/step - loss: 0.0324 - accuracy: 0.9919\n"
     ]
    }
   ],
   "source": [
    "clear_session()\n",
    "history = model.fit(x_train_final, y_train_final, batch_size=batch_size, shuffle=False, epochs=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "4y_3PMV9mAB5"
   },
   "outputs": [],
   "source": [
    "#model.save(DATA_PATH + '/model.keras')\n",
    "#pickle.dump(history, open('history.pkl','wb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "Q1E1eTKwlhi1"
   },
   "outputs": [],
   "source": [
    "#model = load_model(DATA_PATH + '/model.keras')\n",
    "#history = pickle.load(open('history.pkl','rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 2939,
     "status": "ok",
     "timestamp": 1632920444596,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "FB-gwQvHitSN",
    "outputId": "6c1fd226-b488-4b2a-d2bd-b87f58f9ffc0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "261/261 - 2s - loss: 0.5081 - accuracy: 0.9526\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.5080830454826355, 0.9526436924934387]"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(x_test_final, y_test_final, batch_size=batch_size, verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "PBlBMxjmzQaB"
   },
   "outputs": [],
   "source": [
    "#pd.DataFrame(np.reshape(model.predict(x_test_final, batch_size=batch_size), newshape=(-1, num_features_y))).to_csv(DATA_PATH + \"/check_accuracy_predict.csv\")\n",
    "#pd.DataFrame(np.reshape(y_test_final, newshape=(-1, num_features_y))).to_csv(DATA_PATH + \"/check_accuracy_test_real.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "jtGAFgkT184U"
   },
   "outputs": [],
   "source": [
    "#model.predict(x_test_final, batch_size=batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 194
    },
    "executionInfo": {
     "elapsed": 30,
     "status": "ok",
     "timestamp": 1632920444600,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "Mhi89bstitSO",
    "outputId": "83da1228-1d6f-4c3e-af48-26424a1d5e42"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f7f19419090>"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.subplot(211)\n",
    "plt.plot(history.history['accuracy'])\n",
    "plt.title('model accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['accuracy'], loc='upper left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "fognt2N8itSS"
   },
   "outputs": [],
   "source": [
    "def plot_confusion_matrix(cm, classes, title='Confusion matrix', cmap=plt.cm.Blues):\n",
    "\n",
    "    cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title, fontsize=25)\n",
    "    #plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=90, fontsize=15)\n",
    "    plt.yticks(tick_marks, classes, fontsize=15)\n",
    "\n",
    "    fmt = '.2f'\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, format(cm[i, j], fmt),\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm[i, j] > thresh else \"black\", fontsize = 14)\n",
    "\n",
    "    plt.ylabel('True label', fontsize=20)\n",
    "    plt.xlabel('Predicted label', fontsize=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 529
    },
    "executionInfo": {
     "elapsed": 3410,
     "status": "ok",
     "timestamp": 1632920447996,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "tSFStcNFitSS",
    "outputId": "cde6e670-7f2e-477b-e282-2452757788b4"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix = confusion_matrix(np.argmax(np.reshape(y_train_final, newshape=(-1, num_features_y)), axis=1), np.argmax(np.reshape(model.predict(x_train_final, batch_size=batch_size), newshape=(-1, num_features_y)),axis=1))\n",
    "plt.figure(figsize=(7,7))\n",
    "plot_confusion_matrix(cnf_matrix, classes=['Normal','Warning','Alarm'], title=\"Confusion matrix\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "TpqWbVIZVCYY"
   },
   "outputs": [],
   "source": [
    "pd.DataFrame(np.argmax(np.reshape(model.predict(x_test_final, batch_size=batch_size), newshape=(-1, num_features_y)),axis=1)).to_csv(DATA_PATH + \"/check_accuracy_predict_argmax.csv\")\n",
    "pd.DataFrame(np.argmax(np.reshape(y_test_final, newshape=(-1, num_features_y)), axis=1)).to_csv(DATA_PATH + \"/check_accuracy_test_real_argmax.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 529
    },
    "executionInfo": {
     "elapsed": 2275,
     "status": "ok",
     "timestamp": 1632920452266,
     "user": {
      "displayName": "Maurizio Vivarelli",
      "photoUrl": "https://lh3.googleusercontent.com/a/default-user=s64",
      "userId": "09337931341966293696"
     },
     "user_tz": -120
    },
    "id": "W8YSsUOSitST",
    "outputId": "4c184820-0664-46d5-9c68-2fb91efbf0d5"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix = confusion_matrix(np.argmax(np.reshape(y_test_final, newshape=(-1, num_features_y)), axis=1), np.argmax(np.reshape(model.predict(x_test_final, batch_size=batch_size), newshape=(-1, num_features_y)),axis=1))\n",
    "plt.figure(figsize=(7,7))\n",
    "plot_confusion_matrix(cnf_matrix, classes=['Normal','Warning','Alarm'], title=\"Confusion matrix\")\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "Remaining_Life_Estimation_model.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}