outlier-methods-credit-card-fraud-detection-dnn-keras.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### DNN-Keras\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# importing the packages\n",
    "import datetime\n",
    "import warnings\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "from datetime import date\n",
    "import sklearn.exceptions\n",
    "from pylab import rcParams\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.manifold import TSNE\n",
    "from sklearn.svm import OneClassSVM\n",
    "from imblearn.over_sampling import SMOTE\n",
    "from sklearn.ensemble import IsolationForest\n",
    "from imblearn.under_sampling import NearMiss \n",
    "from sklearn.metrics import confusion_matrix,accuracy_score, precision_score, f1_score, recall_score\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.neighbors import LocalOutlierFactor\n",
    "from sklearn.metrics import plot_confusion_matrix\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.metrics import classification_report,accuracy_score\n",
    "warnings.filterwarnings(\"ignore\", category=sklearn.exceptions.UndefinedMetricWarning)\n",
    "\n",
    "\n",
    "from numpy import asarray\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.ensemble import IsolationForest\n",
    "from sklearn.neighbors import LocalOutlierFactor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Dropout\n",
    "import os\n",
    "import matplotlib.pyplot as plt\n",
    "import itertools\n",
    "import seaborn as sn\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import confusion_matrix, accuracy_score, precision_score, f1_score, recall_score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Preprocessing steps"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(129668, 23) (27786, 23)\n"
     ]
    }
   ],
   "source": [
    "data_train = pd.read_csv(\"fraudTrain.csv\")\n",
    "data_test = pd.read_csv(\"fraudTest.csv\")\n",
    "# taking smaller sample to run the model faster\n",
    "\n",
    "df_train= data_train.sample(frac = 0.1,random_state=1)\n",
    "df_test= data_test.sample(frac = 0.05,random_state=1)\n",
    "print(df_train.shape,df_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def getdatafiles():\n",
    "    # loading data\n",
    "    data_train = pd.read_csv(\"fraudTrain.csv\")\n",
    "    data_test = pd.read_csv(\"fraudTest.csv\")\n",
    "    # taking smaller sample to run the model faster\n",
    "\n",
    "    df_train= data_train.sample(frac = 0.1,random_state=1)\n",
    "    df_test= data_test.sample(frac = 0.05,random_state=1)\n",
    "    print(df_train.shape,df_test.shape)\n",
    "    \n",
    "    # function to drop tbe columns\n",
    "    def dropCol(data):\n",
    "        col_to_drop = ['trans_date_trans_time','Unnamed: 0',\n",
    "                       'cc_num','first','last','trans_num','street','unix_time']\n",
    "        res = data.drop(col_to_drop,axis = 1)\n",
    "        return res\n",
    "    # dropping the columns\n",
    "    # dropping the columns ['trans_date_trans_time','Unnamed: 0','cc_num','first','last','trans_num']\n",
    "    # train data set\n",
    "    df_train = dropCol(df_train)\n",
    "    # test data set\n",
    "    df_test = dropCol(df_test)\n",
    "    \n",
    "#     # creating a new feature address\n",
    "#     df_train['add'] = df_train['city'] + df_train['state'] + df_train['zip'].astype(str)\n",
    "#     df_test['add'] = df_test['city'] + df_test['state'] + df_test['zip'].astype(str)\n",
    "#     df_train = df_train.drop(['city','state','zip'],axis =1)\n",
    "#     df_test = df_test.drop(['city','state','zip'],axis =1)\n",
    "#     print ( df_train.shape, df_test.shape)\n",
    "    \n",
    "\n",
    "    #Create independent and Dependent Features\n",
    "    columns = df_train.columns.tolist()\n",
    "\n",
    "    # removing the dependent feature is_fraud\n",
    "    columns = [c for c in columns if c not in [\"is_fraud\"]]\n",
    "\n",
    "    X_train = df_train[columns]\n",
    "    Y_train = df_train['is_fraud']\n",
    "    X_test = df_test[columns]\n",
    "    Y_test = df_test['is_fraud']\n",
    "    print ( X_train.shape, Y_train.shape,X_test.shape, Y_test.shape)\n",
    "\n",
    "    # function to convert dob to years\n",
    "    def age_years(born):\n",
    "        return 2019 - int(born[0:4])\n",
    "\n",
    "    # replacing the dob column with age column in our data set for test and train\n",
    "    X_train['age'] = X_train['dob'].apply(lambda x: age_years(x))\n",
    "    X_train = X_train.drop(['dob'],axis =1)\n",
    "\n",
    "    X_test['age'] = X_test['dob'].apply(lambda x: age_years(x))\n",
    "    X_test = X_test.drop(['dob'],axis =1)\n",
    "    \n",
    "    print(X_train.shape,X_test.shape)\n",
    "\n",
    "    # concanating the test and train data so that number of columns remain the same in both the data sets\n",
    "    final_df = pd.concat([X_train,X_test],axis=0)\n",
    "    final_df.shape\n",
    "\n",
    "\n",
    "    # creating the list of categorical variables\n",
    "    categorical_features =[feature for feature in X_train.columns if final_df[feature].dtypes == 'O']\n",
    "    categorical_features\n",
    "    \n",
    "    # function to convert categorical variables to one hot encoding\n",
    "    def category_onehot_multcols(data,multcolumns):\n",
    "        df_final = data\n",
    "        i=0\n",
    "        for fields in multcolumns:\n",
    "            print(fields)\n",
    "            df1=pd.get_dummies(final_df[fields],drop_first=True)\n",
    "            final_df.drop([fields],axis=1,inplace=True)\n",
    "            if i==0:\n",
    "                df_final=df1.copy()\n",
    "            else:           \n",
    "                df_final=pd.concat([df_final,df1],axis=1)\n",
    "            i=i+1             \n",
    "        df_final=pd.concat([final_df,df_final],axis=1)\n",
    "        return df_final\n",
    "\n",
    "    # applying the one hot encoding\n",
    "    final_df = category_onehot_multcols(final_df, categorical_features)\n",
    "\n",
    "    # removing duplicated columns\n",
    "    final_df =final_df.loc[:,~final_df.columns.duplicated()]\n",
    "    final_df.shape\n",
    "\n",
    "    # separating the test and training data\n",
    "    df_Train=final_df.iloc[:129668,:]\n",
    "    df_Test=final_df.iloc[129668:,:]\n",
    "    print(df_Train.shape,df_Test.shape)\n",
    "    print(Y_train.value_counts(),\"\\n\",Y_test.value_counts())\n",
    "    \n",
    "    # define standard scaler\n",
    "    scaler = StandardScaler()\n",
    "    # transform data\n",
    "    df_Train = scaler.fit_transform(df_Train)\n",
    "    df_Test = scaler.fit_transform(df_Test)# define standard scaler\n",
    "    \n",
    "    \n",
    "    \n",
    "    # files ready for testing on models\n",
    "    print(df_Train.shape, df_Test.shape, Y_train.shape, Y_test.shape)\n",
    "    return df_Train, df_Test, Y_train, Y_test\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# function to evaluate our model \n",
    "def print_eval(y_pred,model):\n",
    "    print(\"Training Accuracy: \",model.score(df_Train, Y_train))\n",
    "    print(\"Testing Accuracy: \", model.score(df_Test, Y_test))\n",
    "    print(confusion_matrix(Y_test, y_pred))\n",
    "    \n",
    "    print(classification_report(Y_test,y_pred))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(129668, 23) (27786, 23)\n",
      "(129668, 14) (129668,) (27786, 14) (27786,)\n",
      "(129668, 14) (27786, 14)\n",
      "merchant\n",
      "category\n",
      "gender\n",
      "city\n",
      "state\n",
      "job\n",
      "(129668, 2120) (27786, 2120)\n",
      "0    128963\n",
      "1       705\n",
      "Name: is_fraud, dtype: int64 \n",
      " 0    27656\n",
      "1      130\n",
      "Name: is_fraud, dtype: int64\n",
      "(129668, 2120) (27786, 2120) (129668,) (27786,)\n"
     ]
    }
   ],
   "source": [
    "df_Train, df_Test, Y_train, Y_test = getdatafiles()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(129668, 2120)\n"
     ]
    }
   ],
   "source": [
    "print(df_Train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "#add input layer\n",
    "model.add(Dense(input_dim = 2120, units = 16, activation = 'relu'))\n",
    "#add 2nd hidden layer\n",
    "model.add(Dense(units = 50, activation = 'relu'))\n",
    "#add dropout layer\n",
    "#model.add(Dropout(0.5))\n",
    "#add 3rd hidden layer\n",
    "model.add(Dense(units = 20, activation = 'relu'))\n",
    "#add 4th hidden layer\n",
    "model.add(Dense(units = 50, activation = 'relu'))\n",
    "#add ouptut layer\n",
    "model.add(Dense(units = 1, activation = 'sigmoid'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_6\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_30 (Dense)             (None, 16)                33936     \n",
      "_________________________________________________________________\n",
      "dense_31 (Dense)             (None, 50)                850       \n",
      "_________________________________________________________________\n",
      "dense_32 (Dense)             (None, 20)                1020      \n",
      "_________________________________________________________________\n",
      "dense_33 (Dense)             (None, 50)                1050      \n",
      "_________________________________________________________________\n",
      "dense_34 (Dense)             (None, 1)                 51        \n",
      "=================================================================\n",
      "Total params: 36,907\n",
      "Trainable params: 36,907\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/150\n",
      "1014/1014 [==============================] - 3s 2ms/step - loss: 0.0919 - accuracy: 0.9804\n",
      "Epoch 2/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0185 - accuracy: 0.9947\n",
      "Epoch 3/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0123 - accuracy: 0.9958\n",
      "Epoch 4/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0079 - accuracy: 0.9972\n",
      "Epoch 5/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0064 - accuracy: 0.9976\n",
      "Epoch 6/150\n",
      "1014/1014 [==============================] - 1s 1000us/step - loss: 0.0044 - accuracy: 0.9983\n",
      "Epoch 7/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0037 - accuracy: 0.9984\n",
      "Epoch 8/150\n",
      "1014/1014 [==============================] - 1s 999us/step - loss: 0.0033 - accuracy: 0.9989\n",
      "Epoch 9/150\n",
      "1014/1014 [==============================] - 1s 1000us/step - loss: 0.0023 - accuracy: 0.9991\n",
      "Epoch 10/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0025 - accuracy: 0.9992\n",
      "Epoch 11/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0024 - accuracy: 0.9991\n",
      "Epoch 12/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0024 - accuracy: 0.9991\n",
      "Epoch 13/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0019 - accuracy: 0.9993\n",
      "Epoch 14/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0016 - accuracy: 0.9994\n",
      "Epoch 15/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0017 - accuracy: 0.9994\n",
      "Epoch 16/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0015 - accuracy: 0.9993\n",
      "Epoch 17/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0014 - accuracy: 0.9994\n",
      "Epoch 18/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0015 - accuracy: 0.9993\n",
      "Epoch 19/150\n",
      "1014/1014 [==============================] - 1s 996us/step - loss: 0.0013 - accuracy: 0.9994\n",
      "Epoch 20/150\n",
      "1014/1014 [==============================] - 1s 997us/step - loss: 0.0011 - accuracy: 0.9994\n",
      "Epoch 21/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 9.4264e-04 - accuracy: 0.9995\n",
      "Epoch 22/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0012 - accuracy: 0.9994\n",
      "Epoch 23/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0011 - accuracy: 0.9994\n",
      "Epoch 24/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0011 - accuracy: 0.9994\n",
      "Epoch 25/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0014 - accuracy: 0.9995\n",
      "Epoch 26/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0011 - accuracy: 0.9994\n",
      "Epoch 27/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0011 - accuracy: 0.9995\n",
      "Epoch 28/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.8166e-04 - accuracy: 0.9995\n",
      "Epoch 29/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 9.3831e-04 - accuracy: 0.9995\n",
      "Epoch 30/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0011 - accuracy: 0.9995\n",
      "Epoch 31/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.7969e-04 - accuracy: 0.9996\n",
      "Epoch 32/150\n",
      "1014/1014 [==============================] - 2s 2ms/step - loss: 9.8008e-04 - accuracy: 0.9995\n",
      "Epoch 33/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.5999e-04 - accuracy: 0.9996\n",
      "Epoch 34/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 9.6193e-04 - accuracy: 0.9995\n",
      "Epoch 35/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 9.6456e-04 - accuracy: 0.9996\n",
      "Epoch 36/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0011 - accuracy: 0.9996\n",
      "Epoch 37/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.2483e-04 - accuracy: 0.9996\n",
      "Epoch 38/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.8643e-04 - accuracy: 0.9996\n",
      "Epoch 39/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 9.0865e-04 - accuracy: 0.9995\n",
      "Epoch 40/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.8205e-04 - accuracy: 0.9995\n",
      "Epoch 41/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.2132e-04 - accuracy: 0.9996: 0s - loss: 7.7\n",
      "Epoch 42/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0011 - accuracy: 0.9994\n",
      "Epoch 43/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0010 - accuracy: 0.9995\n",
      "Epoch 44/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.2528e-04 - accuracy: 0.9996\n",
      "Epoch 45/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.5311e-04 - accuracy: 0.9997\n",
      "Epoch 46/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.5120e-04 - accuracy: 0.9995\n",
      "Epoch 47/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.4756e-04 - accuracy: 0.9996\n",
      "Epoch 48/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.5224e-04 - accuracy: 0.9997\n",
      "Epoch 49/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 9.3743e-04 - accuracy: 0.9996\n",
      "Epoch 50/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.6147e-04 - accuracy: 0.9996\n",
      "Epoch 51/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.8032e-04 - accuracy: 0.9997\n",
      "Epoch 52/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.7899e-04 - accuracy: 0.9996\n",
      "Epoch 53/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.4926e-04 - accuracy: 0.9995\n",
      "Epoch 54/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.6492e-04 - accuracy: 0.9996\n",
      "Epoch 55/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.1740e-04 - accuracy: 0.9997\n",
      "Epoch 56/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.1915e-04 - accuracy: 0.9996\n",
      "Epoch 57/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.6707e-04 - accuracy: 0.9996\n",
      "Epoch 58/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.6167e-04 - accuracy: 0.9997\n",
      "Epoch 59/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.4161e-04 - accuracy: 0.9997\n",
      "Epoch 60/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.9604e-04 - accuracy: 0.9996\n",
      "Epoch 61/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.2928e-04 - accuracy: 0.9996\n",
      "Epoch 62/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.4069e-04 - accuracy: 0.9995\n",
      "Epoch 63/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.8865e-04 - accuracy: 0.9997\n",
      "Epoch 64/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.0520e-04 - accuracy: 0.9995\n",
      "Epoch 65/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.0900e-04 - accuracy: 0.9996\n",
      "Epoch 66/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.8584e-04 - accuracy: 0.9995\n",
      "Epoch 67/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.9458e-04 - accuracy: 0.9997\n",
      "Epoch 68/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.3459e-04 - accuracy: 0.9996\n",
      "Epoch 69/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.4355e-04 - accuracy: 0.9996\n",
      "Epoch 70/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.1251e-04 - accuracy: 0.9996\n",
      "Epoch 71/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.8341e-04 - accuracy: 0.9996\n",
      "Epoch 72/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.3593e-04 - accuracy: 0.9996\n",
      "Epoch 73/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.3709e-04 - accuracy: 0.9996\n",
      "Epoch 74/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.5826e-04 - accuracy: 0.9997\n",
      "Epoch 75/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.0678e-04 - accuracy: 0.9997\n",
      "Epoch 76/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.1703e-04 - accuracy: 0.9997\n",
      "Epoch 77/150\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.9267e-04 - accuracy: 0.9997\n",
      "Epoch 78/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.2288e-04 - accuracy: 0.9997\n",
      "Epoch 79/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.7966e-04 - accuracy: 0.9997\n",
      "Epoch 80/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.2118e-04 - accuracy: 0.9997\n",
      "Epoch 81/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.3875e-04 - accuracy: 0.9997\n",
      "Epoch 82/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.3468e-04 - accuracy: 0.9997\n",
      "Epoch 83/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.4151e-04 - accuracy: 0.9996\n",
      "Epoch 84/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.7797e-04 - accuracy: 0.9997\n",
      "Epoch 85/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.7075e-04 - accuracy: 0.9997\n",
      "Epoch 86/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.1677e-04 - accuracy: 0.9997\n",
      "Epoch 87/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.6856e-04 - accuracy: 0.9997\n",
      "Epoch 88/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.6135e-04 - accuracy: 0.9998\n",
      "Epoch 89/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.4281e-04 - accuracy: 0.9996\n",
      "Epoch 90/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.8204e-04 - accuracy: 0.9996\n",
      "Epoch 91/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.3041e-04 - accuracy: 0.9996\n",
      "Epoch 92/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.4655e-04 - accuracy: 0.9996\n",
      "Epoch 93/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.9564e-04 - accuracy: 0.9997\n",
      "Epoch 94/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 9.0279e-04 - accuracy: 0.9997\n",
      "Epoch 95/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.0655e-04 - accuracy: 0.9997\n",
      "Epoch 96/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0012 - accuracy: 0.9996\n",
      "Epoch 97/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.9562e-04 - accuracy: 0.9997\n",
      "Epoch 98/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.3032e-04 - accuracy: 0.9997\n",
      "Epoch 99/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.8566e-04 - accuracy: 0.9997\n",
      "Epoch 100/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.7401e-04 - accuracy: 0.9998\n",
      "Epoch 101/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.6615e-04 - accuracy: 0.9996\n",
      "Epoch 102/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.4884e-04 - accuracy: 0.9997\n",
      "Epoch 103/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.3810e-04 - accuracy: 0.9998\n",
      "Epoch 104/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.9931e-04 - accuracy: 0.9996\n",
      "Epoch 105/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.7441e-04 - accuracy: 0.9997\n",
      "Epoch 106/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.8001e-04 - accuracy: 0.9997\n",
      "Epoch 107/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.9292e-04 - accuracy: 0.9997\n",
      "Epoch 108/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.5630e-04 - accuracy: 0.9997\n",
      "Epoch 109/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 8.3396e-04 - accuracy: 0.9996\n",
      "Epoch 110/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.7500e-04 - accuracy: 0.9997\n",
      "Epoch 111/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.2010e-04 - accuracy: 0.9997\n",
      "Epoch 112/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.7195e-04 - accuracy: 0.9996\n",
      "Epoch 113/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.9669e-04 - accuracy: 0.9997\n",
      "Epoch 114/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.0494e-04 - accuracy: 0.9997\n",
      "Epoch 115/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.7308e-04 - accuracy: 0.9997\n",
      "Epoch 116/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.9326e-04 - accuracy: 0.9996: 0s - loss: 5.1838e-04 - \n",
      "Epoch 117/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.7242e-04 - accuracy: 0.9997\n",
      "Epoch 118/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.6187e-04 - accuracy: 0.9996\n",
      "Epoch 119/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.5850e-04 - accuracy: 0.9997\n",
      "Epoch 120/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.5424e-04 - accuracy: 0.9997\n",
      "Epoch 121/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.4383e-04 - accuracy: 0.9997\n",
      "Epoch 122/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.7125e-04 - accuracy: 0.9997\n",
      "Epoch 123/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.0529e-04 - accuracy: 0.9996\n",
      "Epoch 124/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0043 - accuracy: 0.9996\n",
      "Epoch 125/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.7571e-04 - accuracy: 0.9996\n",
      "Epoch 126/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.6705e-04 - accuracy: 0.9997\n",
      "Epoch 127/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.8667e-04 - accuracy: 0.9997\n",
      "Epoch 128/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.1174e-04 - accuracy: 0.9997\n",
      "Epoch 129/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.4841e-04 - accuracy: 0.9997\n",
      "Epoch 130/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.5901e-04 - accuracy: 0.9997\n",
      "Epoch 131/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.1064e-04 - accuracy: 0.9998\n",
      "Epoch 132/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 7.1868e-04 - accuracy: 0.9996\n",
      "Epoch 133/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.7582e-04 - accuracy: 0.9996\n",
      "Epoch 134/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0013 - accuracy: 0.9996\n",
      "Epoch 135/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.8896e-04 - accuracy: 0.9997\n",
      "Epoch 136/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.9315e-04 - accuracy: 0.9998\n",
      "Epoch 137/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.0823e-04 - accuracy: 0.9996\n",
      "Epoch 138/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.2321e-04 - accuracy: 0.9997\n",
      "Epoch 139/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.6831e-04 - accuracy: 0.9997\n",
      "Epoch 140/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.9953e-04 - accuracy: 0.9996\n",
      "Epoch 141/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.8254e-04 - accuracy: 0.9997\n",
      "Epoch 142/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.0011e-04 - accuracy: 0.9995\n",
      "Epoch 143/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 6.0923e-04 - accuracy: 0.9997\n",
      "Epoch 144/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 0.0011 - accuracy: 0.9997\n",
      "Epoch 145/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.4481e-04 - accuracy: 0.9997\n",
      "Epoch 146/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 3.8507e-04 - accuracy: 0.9997\n",
      "Epoch 147/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.4244e-04 - accuracy: 0.9997\n",
      "Epoch 148/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.6652e-04 - accuracy: 0.9997\n",
      "Epoch 149/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 5.5133e-04 - accuracy: 0.9997\n",
      "Epoch 150/150\n",
      "1014/1014 [==============================] - 1s 1ms/step - loss: 4.4046e-04 - accuracy: 0.9997\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7fa7b8ee93d0>"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])\n",
    "model.fit(df_Train, Y_train, batch_size = 128, epochs = 150)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "869/869 [==============================] - 1s 585us/step - loss: 0.1671 - accuracy: 0.9910\n",
      "[0.16711823642253876, 0.9910386800765991]\n"
     ]
    }
   ],
   "source": [
    "score = model.evaluate(df_Test, Y_test)\n",
    "print(score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "y_pred = model.predict(df_Test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[27503   153]\n",
      " [   96    34]]\n"
     ]
    }
   ],
   "source": [
    "cm = confusion_matrix(Y_test, y_pred.round())\n",
    "print(cm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.9910386525588426\n",
      "0.18181818181818182\n",
      "0.26153846153846155\n",
      "0.21451104100946375\n"
     ]
    }
   ],
   "source": [
    "print(accuracy_score(Y_test, y_pred.round()))\n",
    "print(precision_score(Y_test, y_pred.round()))\n",
    "print(recall_score(Y_test, y_pred.round()))\n",
    "print(f1_score(Y_test, y_pred.round()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}