aeed_BATADAL.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import modules "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# numpy stack\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "\n",
    "# keras callbacks\n",
    "# from keras.layers import Input, Dense\n",
    "# from keras.models import Model, load_model\n",
    "# from keras import optimizers\n",
    "from keras.callbacks import *\n",
    "\n",
    "# sklearn\n",
    "from sklearn.model_selection import train_test_split\n",
    "# from sklearn.metrics import mean_squared_error\n",
    "from sklearn.metrics import accuracy_score, f1_score, confusion_matrix, auc, precision_score, recall_score\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "# other\n",
    "import time\n",
    "\n",
    "\n",
    "# AutoEncoders for Event Detection (AEED)\n",
    "from aeed import AEED, load_AEED"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load and preprocess training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# load training dataset\n",
    "data_path = \"./data/\"\n",
    "df_train_orig = pd.read_csv(data_path + \"dataset03.csv\", parse_dates = ['DATETIME'], dayfirst=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# get dates and columns with sensor readings\n",
    "dates_train = df_train_orig['DATETIME']\n",
    "sensor_cols = [col for col in df_train_orig.columns if col not in ['DATETIME','ATT_FLAG']]\n",
    "\n",
    "# scale sensor data\n",
    "scaler = MinMaxScaler()\n",
    "X = pd.DataFrame(index = df_train_orig.index, columns = sensor_cols, data = scaler.fit_transform(df_train_orig[sensor_cols]))\n",
    "\n",
    "# split into training and validation\n",
    "X1, X2, _, _  = train_test_split(X, X, test_size=0.33, random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train autoencoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Created autoencoder with structure:\n",
      "layer_0: 43, layer_1: 34, layer_2: 25, layer_3: 17, layer_4: 25, layer_5: 34, layer_6: 43\n"
     ]
    }
   ],
   "source": [
    "# define model parameters\n",
    "params = {\n",
    "    'nI' : X.shape[1],\n",
    "    'nH' : 3,\n",
    "    'cf' : 2.5,\n",
    "    'activation' : 'tanh',\n",
    "    'verbose' : 1,\n",
    "}\n",
    "\n",
    "# create AutoEncoder for Event Detection (AEED)\n",
    "autoencoder = AEED(**params)\n",
    "autoencoder.initialize()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 5869 samples, validate on 2892 samples\n",
      "Epoch 1/500\n",
      " - 5s - loss: 0.0925 - val_loss: 0.0370\n",
      "Epoch 2/500\n",
      " - 1s - loss: 0.0225 - val_loss: 0.0134\n",
      "Epoch 3/500\n",
      " - 1s - loss: 0.0116 - val_loss: 0.0095\n",
      "Epoch 4/500\n",
      " - 1s - loss: 0.0089 - val_loss: 0.0077\n",
      "Epoch 5/500\n",
      " - 1s - loss: 0.0072 - val_loss: 0.0063\n",
      "Epoch 6/500\n",
      " - 1s - loss: 0.0056 - val_loss: 0.0049\n",
      "Epoch 7/500\n",
      " - 1s - loss: 0.0045 - val_loss: 0.0041\n",
      "Epoch 8/500\n",
      " - 1s - loss: 0.0038 - val_loss: 0.0035\n",
      "Epoch 9/500\n",
      " - 1s - loss: 0.0032 - val_loss: 0.0029\n",
      "Epoch 10/500\n",
      " - 1s - loss: 0.0026 - val_loss: 0.0024\n",
      "Epoch 11/500\n",
      " - 1s - loss: 0.0022 - val_loss: 0.0020\n",
      "Epoch 12/500\n",
      " - 1s - loss: 0.0019 - val_loss: 0.0018\n",
      "Epoch 13/500\n",
      " - 1s - loss: 0.0017 - val_loss: 0.0016\n",
      "Epoch 14/500\n",
      " - 1s - loss: 0.0016 - val_loss: 0.0015\n",
      "Epoch 15/500\n",
      " - 1s - loss: 0.0015 - val_loss: 0.0015\n",
      "Epoch 16/500\n",
      " - 1s - loss: 0.0015 - val_loss: 0.0014\n",
      "Epoch 17/500\n",
      " - 1s - loss: 0.0014 - val_loss: 0.0013\n",
      "Epoch 18/500\n",
      " - 1s - loss: 0.0013 - val_loss: 0.0013\n",
      "Epoch 19/500\n",
      " - 1s - loss: 0.0013 - val_loss: 0.0012\n",
      "Epoch 20/500\n",
      " - 1s - loss: 0.0012 - val_loss: 0.0012\n",
      "Epoch 21/500\n",
      " - 1s - loss: 0.0011 - val_loss: 0.0011\n",
      "Epoch 22/500\n",
      " - 1s - loss: 0.0011 - val_loss: 0.0011\n",
      "Epoch 23/500\n",
      " - 1s - loss: 0.0011 - val_loss: 0.0011\n",
      "Epoch 24/500\n",
      " - 1s - loss: 0.0011 - val_loss: 0.0011\n"
     ]
    }
   ],
   "source": [
    "# train models with early stopping and reduction of learning rate on plateau\n",
    "earlyStopping= EarlyStopping(monitor='val_loss', patience=3, verbose=0,  min_delta=1e-4, mode='auto')\n",
    "lr_reduced = ReduceLROnPlateau(monitor='val_loss', factor=0.5, patience=1, verbose=0, epsilon=1e-4, mode='min')\n",
    "    \n",
    "# initialize time\n",
    "start_time = time.time()\n",
    "\n",
    "# train autoencoder\n",
    "autoencoder.train(X1.values,\n",
    "            epochs=500,\n",
    "            batch_size=32,\n",
    "            shuffle=False,\n",
    "            callbacks = [earlyStopping, lr_reduced],\n",
    "            verbose = 2,\n",
    "            validation_data=(X2.values, X2.values))\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test autoencoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# assess detection\n",
    "def compute_scores(Y,Yhat):\n",
    "    return [accuracy_score(Y,Yhat),f1_score(Y,Yhat),precision_score(Y,Yhat),recall_score(Y,Yhat)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Load dataset with attacks\n",
    "df_test_01 = pd.read_csv(data_path + \"dataset04.csv\", parse_dates = ['DATETIME'], dayfirst=True)\n",
    "df_test_02 = pd.read_csv(data_path + \"test_dataset.csv\", parse_dates = ['DATETIME'], dayfirst=True)\n",
    "\n",
    "# scale datasets\n",
    "X3 = pd.DataFrame(index = df_test_01.index, columns = sensor_cols, \n",
    "                  data = scaler.transform(df_test_01[sensor_cols]))\n",
    "X4 = pd.DataFrame(index = df_test_02.index, columns = sensor_cols, \n",
    "                  data = scaler.transform(df_test_02[sensor_cols]))\n",
    "\n",
    "# get targets\n",
    "Y3 = df_test_01['ATT_FLAG']\n",
    "Y4 = df_test_02['ATT_FLAG']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAekAAAEICAYAAAB71gywAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x27c8edcf668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# perform detection\n",
    "\n",
    "# get validation reconstruction errors\n",
    "_, validation_errors = autoencoder.predict(X2)\n",
    "\n",
    "# plot distribution of average validation reconstruction errors \n",
    "f, ax = plt.subplots(1, figsize = (8,4))\n",
    "sns.boxplot(validation_errors.mean(axis=1), ax=ax)\n",
    "ax.set_xlim([0,0.005])\n",
    "ax.set_title('Boxplot of average validation reconstruction errors')\n",
    "\n",
    "# set treshold as quantile of average reconstruction error\n",
    "theta = validation_errors.mean(axis = 1).quantile(0.995)\n",
    "\n",
    "Yhat3, _ = autoencoder.detect(X3, theta = theta , window = 3, average=True)\n",
    "Yhat4, _ = autoencoder.detect(X4, theta = theta, window = 3, average=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Results:\n",
      "\n",
      "                 accuracy  f1_score precision    recall\n",
      "test dataset 01  0.941585  0.730684  0.799517  0.672764\n",
      "test dataset 02  0.923887   0.77886  0.897436  0.687961\n"
     ]
    }
   ],
   "source": [
    "results = pd.DataFrame(index = ['test dataset 01','test dataset 02'], \n",
    "                       columns = ['accuracy','f1_score','precision','recall'])\n",
    "results.loc['test dataset 01'] = compute_scores(Y3,Yhat3)\n",
    "results.loc['test dataset 02'] = compute_scores(Y4,Yhat4)\n",
    "\n",
    "print('Results:\\n')\n",
    "print(results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[Text(0,0,'NO ATTACK'), Text(0,0,'ATTACK')]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABK4AAAHjCAYAAAAKZFdqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x27d5aea7a90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot figures\n",
    "shade_of_gray = '0.75'\n",
    "f, axes = plt.subplots(2,figsize = (20,8))\n",
    "axes[0].plot(Yhat3, color = shade_of_gray, label = 'predicted state')\n",
    "axes[0].fill_between(Yhat3.index, Yhat3.values, where=Yhat3.values <=1, interpolate=True, color=shade_of_gray)\n",
    "axes[0].plot(Y3, color = 'r', alpha = 0.85, lw = 5, label = 'real state')\n",
    "axes[0].set_title('Detection trajectory on test dataset 01', fontsize = 14)\n",
    "axes[0].set_yticks([0,1])\n",
    "axes[0].set_yticklabels(['NO ATTACK','ATTACK'])\n",
    "axes[0].legend(fontsize = 12, loc = 2)\n",
    "\n",
    "axes[1].plot(Yhat4, color = shade_of_gray, label = 'predicted state')\n",
    "axes[1].fill_between(Yhat4.index, Yhat4.values, where=Yhat4.values <=1, interpolate=True, color=shade_of_gray)\n",
    "axes[1].plot(Y4, color = 'r', alpha = 0.75, lw = 5, label = 'real state')\n",
    "axes[1].set_title('Detection trajectory on test dataset 02', fontsize = 14)\n",
    "axes[1].set_yticks([0,1])\n",
    "axes[1].set_yticklabels(['NO ATTACK','ATTACK'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saved AEED parameters to autoencoder.json.\n",
      "Keras model saved to autoencoder.h5\n"
     ]
    }
   ],
   "source": [
    "# save autoencoder\n",
    "autoencoder.save('autoencoder')\n",
    "del autoencoder"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compare ROC curves"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# load autoencoder\n",
    "autoencoder = load_AEED('autoencoder.json','autoencoder.h5')\n",
    "\n",
    "# get validation reconstruction errors\n",
    "_, validation_errors = autoencoder.predict(X2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "def compute_roc_curve(X, Y, window, quantiles, verbose = False):    \n",
    "    print('Computing ROC curve for window = %d'%(window))    \n",
    "    df_roc = pd.DataFrame(index = quantiles, columns = ['fpr','tpr'])\n",
    "    for qtile in quantiles:\n",
    "        # compute results\n",
    "        theta = validation_errors.mean(axis = 1).quantile(qtile)\n",
    "        if verbose:\n",
    "            print('Quantile: {0:.3f}, threshold value: {1:.5}'.format(qtile,theta))\n",
    "        Yhat, _ = autoencoder.detect(X, theta = theta, window = window, average=True)\n",
    "        CM = confusion_matrix(Y,Yhat)\n",
    "        TN = CM[0,0]\n",
    "        FP = CM[0,1]\n",
    "        FN = CM[1,0]\n",
    "        TP = CM[1,1]        \n",
    "        fpr = FP/(FP+TN)\n",
    "        tpr = TP/(TP+FN) # or recall_score, same stuff\n",
    "        df_roc.loc[qtile,:] = [fpr, tpr]\n",
    "\n",
    "    # compute auc\n",
    "    df_roc.loc[0,:] = (1,1)\n",
    "    df_roc.loc[1000,:] = (0,0)    \n",
    "    area = auc(df_roc['fpr'],df_roc['tpr'], reorder =True)\n",
    "        \n",
    "    return df_roc.sort_index(), area\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Computing ROC curve for window = 1\n",
      "Computing ROC curve for window = 3\n",
      "Computing ROC curve for window = 6\n",
      "Computing ROC curve for window = 12\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x27d70a57c88>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAlYAAAJCCAYAAAAC4omSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nD6w4juM4juN0hAdWHMdxHMdxOvL/JqTE8nJ6zI8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x27d70a46ef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "windows = [1,3,6,12]\n",
    "colors = ['blue', 'green', 'black', 'red']\n",
    "markers = ['s','o','x','v']\n",
    "\n",
    "thresholds = np.linspace(0.8,1,51)\n",
    "f, ax = plt.subplots(1,figsize=(10,10))\n",
    "for window, color, marker in zip(windows,colors,markers):\n",
    "    df_roc, area = compute_roc_curve(X3, Y3, window, thresholds)\n",
    "    ax.plot(df_roc.fpr, df_roc.tpr,color = color, lw = 0, linestyle = None, \n",
    "        marker = marker, markersize = 10, markerfacecolor = color, \n",
    "            label='Window: {0:d} hours, AUC: {1:.3f}'.format(window,area))\n",
    "ax.legend(fontsize = 15)\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}