bank_full_Ensample_Analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "smart-feeding",
   "metadata": {},
   "source": [
    "## bank-full_data.csv (Ensemble Techniques and DNN)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "healthy-password",
   "metadata": {},
   "source": [
    "Data Description:\n",
    "The data is related with direct marketing campaigns of a Portuguese banking institution. \n",
    "The marketing campaigns were based on phone calls. \n",
    "Often, more than one contact to the same client was required, in order to access if the product (bank term deposit) \n",
    "would be ('yes') or not ('no') subscribed."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "adverse-stake",
   "metadata": {},
   "source": [
    "To begin this exploratory analysis, first import libraries. \n",
    "In this notebook you will learn \n",
    "1. Basic EDA (Expolary Data Analysis)\n",
    "2. How to use matplotlib effectively \n",
    "3. How to handle missing values\n",
    "4. How to apply ensample techniques like XGBoosting\n",
    "5. how to tune XGBoost parameters\n",
    "6. Feature importance graph\n",
    "7. build basic ANN\n",
    "8. Tune the ANN parameters such as learning rate, activation function, optimizer "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "greater-sapphire",
   "metadata": {},
   "outputs": [],
   "source": [
    "#import librariesimport pandas as pd\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "import sklearn\n",
    "import warnings \n",
    "warnings.filterwarnings('ignore')\n",
    "import xgboost as xgb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "reflected-bracelet",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "age           int64\n",
       "job          object\n",
       "marital      object\n",
       "education    object\n",
       "default      object\n",
       "balance       int64\n",
       "housing      object\n",
       "loan         object\n",
       "contact      object\n",
       "day           int64\n",
       "month        object\n",
       "duration      int64\n",
       "campaign      int64\n",
       "pdays         int64\n",
       "previous      int64\n",
       "poutcome     object\n",
       "y            object\n",
       "dtype: object"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#importing dataset\n",
    "df= pd.read_csv(\"bank-full.csv\", sep = \";\")\n",
    "df.dtypes\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "threaded-cathedral",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 45211 entries, 0 to 45210\n",
      "Data columns (total 17 columns):\n",
      " #   Column     Non-Null Count  Dtype \n",
      "---  ------     --------------  ----- \n",
      " 0   age        45211 non-null  int64 \n",
      " 1   job        45211 non-null  object\n",
      " 2   marital    45211 non-null  object\n",
      " 3   education  45211 non-null  object\n",
      " 4   default    45211 non-null  object\n",
      " 5   balance    45211 non-null  int64 \n",
      " 6   housing    45211 non-null  object\n",
      " 7   loan       45211 non-null  object\n",
      " 8   contact    45211 non-null  object\n",
      " 9   day        45211 non-null  int64 \n",
      " 10  month      45211 non-null  object\n",
      " 11  duration   45211 non-null  int64 \n",
      " 12  campaign   45211 non-null  int64 \n",
      " 13  pdays      45211 non-null  int64 \n",
      " 14  previous   45211 non-null  int64 \n",
      " 15  poutcome   45211 non-null  object\n",
      " 16  y          45211 non-null  object\n",
      "dtypes: int64(7), object(10)\n",
      "memory usage: 5.9+ MB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "standard-logan",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "age          0\n",
       "job          0\n",
       "marital      0\n",
       "education    0\n",
       "default      0\n",
       "balance      0\n",
       "housing      0\n",
       "loan         0\n",
       "contact      0\n",
       "day          0\n",
       "month        0\n",
       "duration     0\n",
       "campaign     0\n",
       "pdays        0\n",
       "previous     0\n",
       "poutcome     0\n",
       "y            0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Check the Null vallue of the daqta set\n",
    "df.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "dependent-celebration",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "blue-collar      9732\n",
       "management       9458\n",
       "technician       7597\n",
       "admin.           5171\n",
       "services         4154\n",
       "retired          2264\n",
       "self-employed    1579\n",
       "entrepreneur     1487\n",
       "unemployed       1303\n",
       "housemaid        1240\n",
       "student           938\n",
       "unknown           288\n",
       "Name: job, dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['job'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "upper-damages",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "cellular     29285\n",
       "unknown      13020\n",
       "telephone     2906\n",
       "Name: contact, dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['contact'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "honey-lightning",
   "metadata": {},
   "source": [
    "So,The contact feature has some unknown values so, it might be null values. So, We have to check all the feature \n",
    "where it has null values. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "welcome-thing",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Columns with Unknown Values are   \n",
      "['job', 'education', 'contact', 'poutcome']\n"
     ]
    }
   ],
   "source": [
    "# So, print the all columns which have unknown values\n",
    "\n",
    "colum_unknown = []\n",
    "for col in df.columns:\n",
    "    if 'unknown' in df[col].values:\n",
    "        colum_unknown.append(col)\n",
    "        \n",
    "print(\"Columns with Unknown Values are  \"+\" \\n\" + str(colum_unknown))      "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "systematic-framing",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Unknown values count : \n",
      "\n",
      "job  :  288\n",
      "education  :  1857\n",
      "contact  :  13020\n",
      "poutcome  :  36959\n"
     ]
    }
   ],
   "source": [
    "# Find the unknoun values of each column which we found abouve\n",
    "print(\"Unknown values count : \\n\")\n",
    "for col in colum_unknown:\n",
    "    print(col,\" : \",df[df[col].str.contains('unknown')][col].count())\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "greatest-slave",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "age            77\n",
       "job            12\n",
       "marital         3\n",
       "education       4\n",
       "default         2\n",
       "balance      7168\n",
       "housing         2\n",
       "loan            2\n",
       "contact         3\n",
       "day            31\n",
       "month          12\n",
       "duration     1573\n",
       "campaign       48\n",
       "pdays         559\n",
       "previous       41\n",
       "poutcome        4\n",
       "y               2\n",
       "dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.apply(lambda x:len(x.unique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "continuing-domain",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Seperate  the dependent variables and indipendent variables\n",
    "y = df.iloc[:, -1].values\n",
    "numeric_df = df._get_numeric_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "motivated-coaching",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['age', 'balance', 'day', 'duration', 'campaign', 'pdays', 'previous'], dtype='object')\n"
     ]
    }
   ],
   "source": [
    "print(numeric_df.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "satisfactory-payday",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 0 0 ... 1 0 0]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.preprocessing import LabelEncoder\n",
    "label=LabelEncoder()\n",
    "y=label.fit_transform(y)\n",
    "print(y)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "undefined-separation",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 720x576 with 0 Axes>"
      ]
     },
     "execution_count": 15,
     "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"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 720x576 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.countplot(df['y'],data=df,palette='spring')\n",
    "plt.title(' Unique values for dependent feature')\n",
    "plt.figure(figsize=(10,8))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "blind-customer",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Split dataset into training and test set\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "numeric_train, numeric_test, y_train, y_test = train_test_split(numeric_df, y,\n",
    "                                                    test_size =0.2,random_state=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "nutritional-warrant",
   "metadata": {},
   "source": [
    "### Understand the xgb.train method then, According to develop the values for each function."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "absent-shell",
   "metadata": {},
   "source": [
    "xgb.train(\n",
    "    params,\n",
    "    dtrain,\n",
    "    num_boost_round=10,\n",
    "    evals=(),\n",
    "    obj=None,\n",
    "    feval=None,\n",
    "    maximize=False,\n",
    "    early_stopping_rounds=None,\n",
    "    evals_result=None,\n",
    "    verbose_eval=True,\n",
    "    xgb_model=None,\n",
    "    callbacks=None)\n",
    "params : dict\n",
    "    Booster params.\n",
    "dtrain : DMatrix\n",
    "    Data to be trained.\n",
    "num_boost_round: int\n",
    "    Number of boosting iterations.\n",
    "evals: list of pairs (DMatrix, string)\n",
    "    List of validation sets for which metrics will evaluated during training.\n",
    "    Validation metrics will help us track the performance of the model.\n",
    "obj : function\n",
    "    Customized objective function.\n",
    "feval : function\n",
    "    Customized evaluation function.\n",
    "maximize : bool\n",
    "    Whether to maximize feval.\n",
    "early_stopping_rounds: int\n",
    "    Activates early stopping. Validation metric needs to improve at least once in\n",
    "    every **early_stopping_rounds** round(s) to continue training.\n",
    "    Requires at least one item in **evals**."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "controlling-village",
   "metadata": {},
   "source": [
    "### Part1: Create the XGBoost model only for numeric variable"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "complicated-latin",
   "metadata": {},
   "source": [
    "1. Create the xgboost matrices\n",
    "2. set the parameters for learning rate, max_depth, colsample_bytree, child_weight etc..\n",
    "3. develop the xgboost model\n",
    "4. Tune the parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "distinct-celtic",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create xgboost matrices\n",
    "\n",
    "Train_1 = xgb.DMatrix(numeric_train, label = y_train)\n",
    "Test_1 = xgb.DMatrix(numeric_test, label = y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "boring-coordination",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set the parameters\n",
    "\n",
    "parameters1 = {'learning_rate': 0.3,\n",
    "               'max_depth': 2,\n",
    "               'colsample_bytree': 1,\n",
    "               'subsample': 1,\n",
    "               'min_child_weight': 1,\n",
    "               'gamma': 0, \n",
    "               'random_state': 1502,\n",
    "               'eval_metric': \"auc\",\n",
    "               'objective': \"binary:logistic\"}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "amber-theater",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0]\tYes-auc:0.76399\n",
      "[50]\tYes-auc:0.87565\n",
      "[100]\tYes-auc:0.88043\n",
      "[150]\tYes-auc:0.88216\n",
      "[199]\tYes-auc:0.88212\n"
     ]
    }
   ],
   "source": [
    "#run XGBoost\n",
    "\n",
    "model1 = xgb.train(params = parameters1,\n",
    "                   dtrain = Train_1,\n",
    "                   num_boost_round = 200,\n",
    "                   evals = [(Test_1, \"Yes\")],\n",
    "                   verbose_eval = 50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "upper-symphony",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Predictions for test data set\n",
    "\n",
    "predictions1 = model1.predict(Test_1)\n",
    "predictions1 = np.where(predictions1 > 0.5, 1, 0)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "growing-vatican",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy of the train data set 0.8952781156695787\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.92      0.97      0.94      7993\n",
      "           1       0.58      0.34      0.43      1050\n",
      "\n",
      "    accuracy                           0.90      9043\n",
      "   macro avg       0.75      0.65      0.69      9043\n",
      "weighted avg       0.88      0.90      0.88      9043\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Confusion Matrix\n",
    "\n",
    "from sklearn.metrics import confusion_matrix,classification_report,accuracy_score\n",
    "from sklearn.metrics import classification_report\n",
    "confusion_matrix1 = confusion_matrix(y_test, predictions1)\n",
    "report1 = classification_report(y_test, predictions1)\n",
    "print(\"accuracy of the train data set\" + \" \"+ str(accuracy_score(y_test,predictions1)))\n",
    "print(report1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "historic-implementation",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(69.0, 0.5, 'Truth')"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,8))\n",
    "sns.heatmap(confusion_matrix1,annot=True,fmt='d')\n",
    "plt.xlabel('Predicted')\n",
    "plt.ylabel('Truth')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "atomic-current",
   "metadata": {},
   "source": [
    "### Part2: Creating  the XGBoost model for numerical variables and categorical variables"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "express-annual",
   "metadata": {},
   "source": [
    " Now consider the catogrical variable also for crating the model and see how accuracy change "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "composed-assembly",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>job</th>\n",
       "      <th>marital</th>\n",
       "      <th>education</th>\n",
       "      <th>default</th>\n",
       "      <th>housing</th>\n",
       "      <th>loan</th>\n",
       "      <th>contact</th>\n",
       "      <th>month</th>\n",
       "      <th>poutcome</th>\n",
       "      <th>y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>management</td>\n",
       "      <td>married</td>\n",
       "      <td>tertiary</td>\n",
       "      <td>no</td>\n",
       "      <td>yes</td>\n",
       "      <td>no</td>\n",
       "      <td>unknown</td>\n",
       "      <td>may</td>\n",
       "      <td>unknown</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>technician</td>\n",
       "      <td>single</td>\n",
       "      <td>secondary</td>\n",
       "      <td>no</td>\n",
       "      <td>yes</td>\n",
       "      <td>no</td>\n",
       "      <td>unknown</td>\n",
       "      <td>may</td>\n",
       "      <td>unknown</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>entrepreneur</td>\n",
       "      <td>married</td>\n",
       "      <td>secondary</td>\n",
       "      <td>no</td>\n",
       "      <td>yes</td>\n",
       "      <td>yes</td>\n",
       "      <td>unknown</td>\n",
       "      <td>may</td>\n",
       "      <td>unknown</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>blue-collar</td>\n",
       "      <td>married</td>\n",
       "      <td>unknown</td>\n",
       "      <td>no</td>\n",
       "      <td>yes</td>\n",
       "      <td>no</td>\n",
       "      <td>unknown</td>\n",
       "      <td>may</td>\n",
       "      <td>unknown</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>unknown</td>\n",
       "      <td>single</td>\n",
       "      <td>unknown</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>unknown</td>\n",
       "      <td>may</td>\n",
       "      <td>unknown</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45206</th>\n",
       "      <td>technician</td>\n",
       "      <td>married</td>\n",
       "      <td>tertiary</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>cellular</td>\n",
       "      <td>nov</td>\n",
       "      <td>unknown</td>\n",
       "      <td>yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45207</th>\n",
       "      <td>retired</td>\n",
       "      <td>divorced</td>\n",
       "      <td>primary</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>cellular</td>\n",
       "      <td>nov</td>\n",
       "      <td>unknown</td>\n",
       "      <td>yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45208</th>\n",
       "      <td>retired</td>\n",
       "      <td>married</td>\n",
       "      <td>secondary</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>cellular</td>\n",
       "      <td>nov</td>\n",
       "      <td>success</td>\n",
       "      <td>yes</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45209</th>\n",
       "      <td>blue-collar</td>\n",
       "      <td>married</td>\n",
       "      <td>secondary</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>telephone</td>\n",
       "      <td>nov</td>\n",
       "      <td>unknown</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45210</th>\n",
       "      <td>entrepreneur</td>\n",
       "      <td>married</td>\n",
       "      <td>secondary</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>no</td>\n",
       "      <td>cellular</td>\n",
       "      <td>nov</td>\n",
       "      <td>other</td>\n",
       "      <td>no</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>45211 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                job   marital  education default housing loan    contact  \\\n",
       "0        management   married   tertiary      no     yes   no    unknown   \n",
       "1        technician    single  secondary      no     yes   no    unknown   \n",
       "2      entrepreneur   married  secondary      no     yes  yes    unknown   \n",
       "3       blue-collar   married    unknown      no     yes   no    unknown   \n",
       "4           unknown    single    unknown      no      no   no    unknown   \n",
       "...             ...       ...        ...     ...     ...  ...        ...   \n",
       "45206    technician   married   tertiary      no      no   no   cellular   \n",
       "45207       retired  divorced    primary      no      no   no   cellular   \n",
       "45208       retired   married  secondary      no      no   no   cellular   \n",
       "45209   blue-collar   married  secondary      no      no   no  telephone   \n",
       "45210  entrepreneur   married  secondary      no      no   no   cellular   \n",
       "\n",
       "      month poutcome    y  \n",
       "0       may  unknown   no  \n",
       "1       may  unknown   no  \n",
       "2       may  unknown   no  \n",
       "3       may  unknown   no  \n",
       "4       may  unknown   no  \n",
       "...     ...      ...  ...  \n",
       "45206   nov  unknown  yes  \n",
       "45207   nov  unknown  yes  \n",
       "45208   nov  success  yes  \n",
       "45209   nov  unknown   no  \n",
       "45210   nov    other   no  \n",
       "\n",
       "[45211 rows x 10 columns]"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# exctract all catergorical data\n",
    "\n",
    "df.select_dtypes(exclude='number')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "competitive-europe",
   "metadata": {},
   "source": [
    "Isolate the all catogrical variables and convert to the numberic values(get into the dummy variables)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "desperate-participant",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['age', 'balance', 'day', 'duration', 'campaign', 'pdays', 'previous', 'job_blue-collar', 'job_entrepreneur', 'job_housemaid', 'job_management', 'job_retired', 'job_self-employed', 'job_services', 'job_student', 'job_technician', 'job_unemployed', 'job_unknown', 'marital_married', 'marital_single', 'education_secondary', 'education_tertiary', 'education_unknown', 'default_yes', 'housing_yes', 'loan_yes', 'contact_telephone', 'contact_unknown', 'month_aug', 'month_dec', 'month_feb', 'month_jan', 'month_jul', 'month_jun', 'month_mar', 'month_may', 'month_nov', 'month_oct', 'month_sep', 'poutcome_other', 'poutcome_success', 'poutcome_unknown']\n"
     ]
    }
   ],
   "source": [
    "#isolate the categorical variables\n",
    "df_categorical = df.select_dtypes(exclude = \"number\")\n",
    "\n",
    "#tranform categorical variables into dummy variables\n",
    "df_categorical = pd.get_dummies(data = df_categorical,\n",
    "                                     drop_first = True)\n",
    "\n",
    "#joining numerical and categorical datasets\n",
    "final_df = pd.concat([numeric_df, df_categorical], axis = 1)\n",
    "#getting names of columns\n",
    "feature_columns = list(final_df.columns.values)\n",
    "#feature_columns = list(final_df.columns)\n",
    "#drop the dependent feature from the final_df\n",
    "feature_columns = feature_columns[:-1]\n",
    "print(feature_columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "id": "enclosed-driving",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>age</th>\n",
       "      <th>balance</th>\n",
       "      <th>day</th>\n",
       "      <th>duration</th>\n",
       "      <th>campaign</th>\n",
       "      <th>pdays</th>\n",
       "      <th>previous</th>\n",
       "      <th>job_blue-collar</th>\n",
       "      <th>job_entrepreneur</th>\n",
       "      <th>job_housemaid</th>\n",
       "      <th>...</th>\n",
       "      <th>month_jun</th>\n",
       "      <th>month_mar</th>\n",
       "      <th>month_may</th>\n",
       "      <th>month_nov</th>\n",
       "      <th>month_oct</th>\n",
       "      <th>month_sep</th>\n",
       "      <th>poutcome_other</th>\n",
       "      <th>poutcome_success</th>\n",
       "      <th>poutcome_unknown</th>\n",
       "      <th>y_yes</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>58</td>\n",
       "      <td>2143</td>\n",
       "      <td>5</td>\n",
       "      <td>261</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>44</td>\n",
       "      <td>29</td>\n",
       "      <td>5</td>\n",
       "      <td>151</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>33</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>76</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>47</td>\n",
       "      <td>1506</td>\n",
       "      <td>5</td>\n",
       "      <td>92</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>33</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>198</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 43 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   age  balance  day  duration  campaign  pdays  previous  job_blue-collar  \\\n",
       "0   58     2143    5       261         1     -1         0                0   \n",
       "1   44       29    5       151         1     -1         0                0   \n",
       "2   33        2    5        76         1     -1         0                0   \n",
       "3   47     1506    5        92         1     -1         0                1   \n",
       "4   33        1    5       198         1     -1         0                0   \n",
       "\n",
       "   job_entrepreneur  job_housemaid  ...  month_jun  month_mar  month_may  \\\n",
       "0                 0              0  ...          0          0          1   \n",
       "1                 0              0  ...          0          0          1   \n",
       "2                 1              0  ...          0          0          1   \n",
       "3                 0              0  ...          0          0          1   \n",
       "4                 0              0  ...          0          0          1   \n",
       "\n",
       "   month_nov  month_oct  month_sep  poutcome_other  poutcome_success  \\\n",
       "0          0          0          0               0                 0   \n",
       "1          0          0          0               0                 0   \n",
       "2          0          0          0               0                 0   \n",
       "3          0          0          0               0                 0   \n",
       "4          0          0          0               0                 0   \n",
       "\n",
       "   poutcome_unknown  y_yes  \n",
       "0                 1      0  \n",
       "1                 1      0  \n",
       "2                 1      0  \n",
       "3                 1      0  \n",
       "4                 1      0  \n",
       "\n",
       "[5 rows x 43 columns]"
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "final_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "id": "correct-bangladesh",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Isolate the x and y variables\n",
    "\n",
    "y2 = final_df.iloc[:, -1].values\n",
    "x2 = final_df.iloc[:, :-1].values\n",
    "    \n",
    "# Split dataset into 80% into training and 20% for testing set\n",
    "x2_train, x2_test, y2_train, y2_test = train_test_split(x2, y2,\n",
    "                                                    test_size =0.2,\n",
    "                                                    random_state=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "id": "provincial-latin",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create xgboost matrices\n",
    "\n",
    "Train_2 = xgb.DMatrix(x2_train, label = y2_train, feature_names = feature_columns)\n",
    "Test_2 = xgb.DMatrix(x2_test, label = y2_test, feature_names = feature_columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "id": "structured-gabriel",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set the parameters\n",
    "parameters2 = {'learning_rate': 0.3,\n",
    "               'max_depth': 2,\n",
    "               'colsample_bytree': 1,\n",
    "               'subsample': 1,\n",
    "               'min_child_weight': 1,\n",
    "               'gamma': 0, \n",
    "               'random_state': 1502,\n",
    "               'eval_metric': \"auc\",\n",
    "               'objective': \"binary:logistic\"}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "id": "retired-spectrum",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0]\tYes-auc:0.75486\n",
      "[50]\tYes-auc:0.92029\n",
      "[100]\tYes-auc:0.92838\n",
      "[150]\tYes-auc:0.92993\n",
      "[199]\tYes-auc:0.93049\n"
     ]
    }
   ],
   "source": [
    "# Run XGBoost\n",
    "model2 = xgb.train(params = parameters2,\n",
    "                   dtrain = Train_2,\n",
    "                   num_boost_round = 200,\n",
    "                   evals = [(Test_2, \"Yes\")],\n",
    "                   verbose_eval = 50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "id": "unnecessary-plain",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy of the train data set 0.9085480482140883\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.93      0.97      0.95      7993\n",
      "           1       0.66      0.44      0.53      1050\n",
      "\n",
      "    accuracy                           0.91      9043\n",
      "   macro avg       0.79      0.71      0.74      9043\n",
      "weighted avg       0.90      0.91      0.90      9043\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Predictions\n",
    "predictions2 = model2.predict(Test_2)\n",
    "predictions2 = np.where(predictions2 > 0.5, 1, 0)\n",
    "\n",
    "# Confusion Matrix\n",
    "confusion_matrix2 = confusion_matrix(y2_test, predictions2)\n",
    "\n",
    "print(\"accuracy of the train data set\" + \" \"+ str(accuracy_score(y2_test,predictions2)))\n",
    "report2 = classification_report(y2_test, predictions2)\n",
    "print(report2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "id": "comparative-conclusion",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(69.0, 0.5, 'Truth')"
      ]
     },
     "execution_count": 125,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,8))\n",
    "sns.heatmap(confusion_matrix2,annot=True,fmt='d')\n",
    "plt.xlabel('Predicted')\n",
    "plt.ylabel('Truth')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "terminal-grade",
   "metadata": {},
   "source": [
    "### XGBoost model with KFold cross-validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "id": "tight-digit",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Setting the cross validation parameters\n",
    "\n",
    "from sklearn.model_selection import KFold\n",
    "tune_control = KFold(n_splits = 5,\n",
    "                     shuffle = True,\n",
    "                     random_state = 1).split(X = x2_train,\n",
    "                                                y = y2_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "recognized-alpha",
   "metadata": {},
   "source": [
    "### Tune the hyper parameters with cross-validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "id": "marked-example",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set parameter tuning\n",
    "\n",
    "tune_grid = {'learning_rate': [0.05, 0.3],\n",
    "               'max_depth': range(2, 9, 2),\n",
    "               'colsample_bytree': [0.5, 1],\n",
    "               'subsample': [1],\n",
    "               'min_child_weight': [1],\n",
    "               'gamma': [0], \n",
    "               'random_state': [1502],\n",
    "               'n_estimators': range(200, 2000, 200),\n",
    "               'booster': [\"gbtree\"]}  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "id": "flush-french",
   "metadata": {},
   "outputs": [],
   "source": [
    "# State that we are doing a classification problem\n",
    "\n",
    "from xgboost import XGBClassifier\n",
    "classifier = XGBClassifier(objective = \"binary:logistic\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "sitting-companion",
   "metadata": {},
   "source": [
    "### Try  First cross validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "id": "fantastic-weather",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Cross Validation Assembly\n",
    "\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "grid_search = GridSearchCV(estimator = classifier,\n",
    "                           param_grid = tune_grid,\n",
    "                            scoring = \"roc_auc\",\n",
    "                            n_jobs = 6,\n",
    "                            cv = tune_control,\n",
    "                            verbose = 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "id": "southern-canal",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Setting evaluation parameters\n",
    "evaluation_parameters = {\"early_stopping_rounds\": 100,\n",
    "                         \"eval_metric\": \"auc\",\n",
    "                         \"eval_set\": [(x2_test, y2_test)]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "id": "rough-knife",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 144 candidates, totalling 720 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=6)]: Using backend LokyBackend with 6 concurrent workers.\n",
      "[Parallel(n_jobs=6)]: Done   6 tasks      | elapsed:   27.3s\n",
      "[Parallel(n_jobs=6)]: Done  60 tasks      | elapsed:  8.2min\n",
      "[Parallel(n_jobs=6)]: Done 150 tasks      | elapsed: 20.6min\n",
      "[Parallel(n_jobs=6)]: Done 276 tasks      | elapsed: 28.4min\n",
      "[Parallel(n_jobs=6)]: Done 438 tasks      | elapsed: 423.7min\n",
      "[Parallel(n_jobs=6)]: Done 636 tasks      | elapsed: 449.2min\n",
      "[Parallel(n_jobs=6)]: Done 720 out of 720 | elapsed: 456.4min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0]\tvalidation_0-auc:0.85513\n",
      "Will train until validation_0-auc hasn't improved in 100 rounds.\n",
      "[1]\tvalidation_0-auc:0.86474\n",
      "[2]\tvalidation_0-auc:0.89285\n",
      "[3]\tvalidation_0-auc:0.89638\n",
      "[4]\tvalidation_0-auc:0.90129\n",
      "[5]\tvalidation_0-auc:0.90248\n",
      "[6]\tvalidation_0-auc:0.90451\n",
      "[7]\tvalidation_0-auc:0.90742\n",
      "[8]\tvalidation_0-auc:0.90887\n",
      "[9]\tvalidation_0-auc:0.90900\n",
      "[10]\tvalidation_0-auc:0.90982\n",
      "[11]\tvalidation_0-auc:0.91043\n",
      "[12]\tvalidation_0-auc:0.91208\n",
      "[13]\tvalidation_0-auc:0.91362\n",
      "[14]\tvalidation_0-auc:0.91408\n",
      "[15]\tvalidation_0-auc:0.91405\n",
      "[16]\tvalidation_0-auc:0.91363\n",
      "[17]\tvalidation_0-auc:0.91388\n",
      "[18]\tvalidation_0-auc:0.91337\n",
      "[19]\tvalidation_0-auc:0.91431\n",
      "[20]\tvalidation_0-auc:0.91438\n",
      "[21]\tvalidation_0-auc:0.91455\n",
      "[22]\tvalidation_0-auc:0.91498\n",
      "[23]\tvalidation_0-auc:0.91568\n",
      "[24]\tvalidation_0-auc:0.91600\n",
      "[25]\tvalidation_0-auc:0.91672\n",
      "[26]\tvalidation_0-auc:0.91724\n",
      "[27]\tvalidation_0-auc:0.91724\n",
      "[28]\tvalidation_0-auc:0.91734\n",
      "[29]\tvalidation_0-auc:0.91810\n",
      "[30]\tvalidation_0-auc:0.91795\n",
      "[31]\tvalidation_0-auc:0.91822\n",
      "[32]\tvalidation_0-auc:0.91835\n",
      "[33]\tvalidation_0-auc:0.91902\n",
      "[34]\tvalidation_0-auc:0.91910\n",
      "[35]\tvalidation_0-auc:0.91968\n",
      "[36]\tvalidation_0-auc:0.91992\n",
      "[37]\tvalidation_0-auc:0.92032\n",
      "[38]\tvalidation_0-auc:0.92059\n",
      "[39]\tvalidation_0-auc:0.92091\n",
      "[40]\tvalidation_0-auc:0.92159\n",
      "[41]\tvalidation_0-auc:0.92173\n",
      "[42]\tvalidation_0-auc:0.92187\n",
      "[43]\tvalidation_0-auc:0.92179\n",
      "[44]\tvalidation_0-auc:0.92191\n",
      "[45]\tvalidation_0-auc:0.92222\n",
      "[46]\tvalidation_0-auc:0.92212\n",
      "[47]\tvalidation_0-auc:0.92245\n",
      "[48]\tvalidation_0-auc:0.92256\n",
      "[49]\tvalidation_0-auc:0.92253\n",
      "[50]\tvalidation_0-auc:0.92280\n",
      "[51]\tvalidation_0-auc:0.92302\n",
      "[52]\tvalidation_0-auc:0.92330\n",
      "[53]\tvalidation_0-auc:0.92344\n",
      "[54]\tvalidation_0-auc:0.92339\n",
      "[55]\tvalidation_0-auc:0.92338\n",
      "[56]\tvalidation_0-auc:0.92333\n",
      "[57]\tvalidation_0-auc:0.92327\n",
      "[58]\tvalidation_0-auc:0.92332\n",
      "[59]\tvalidation_0-auc:0.92336\n",
      "[60]\tvalidation_0-auc:0.92326\n",
      "[61]\tvalidation_0-auc:0.92307\n",
      "[62]\tvalidation_0-auc:0.92348\n",
      "[63]\tvalidation_0-auc:0.92359\n",
      "[64]\tvalidation_0-auc:0.92358\n",
      "[65]\tvalidation_0-auc:0.92395\n",
      "[66]\tvalidation_0-auc:0.92383\n",
      "[67]\tvalidation_0-auc:0.92399\n",
      "[68]\tvalidation_0-auc:0.92413\n",
      "[69]\tvalidation_0-auc:0.92395\n",
      "[70]\tvalidation_0-auc:0.92379\n",
      "[71]\tvalidation_0-auc:0.92390\n",
      "[72]\tvalidation_0-auc:0.92397\n",
      "[73]\tvalidation_0-auc:0.92415\n",
      "[74]\tvalidation_0-auc:0.92415\n",
      "[75]\tvalidation_0-auc:0.92420\n",
      "[76]\tvalidation_0-auc:0.92443\n",
      "[77]\tvalidation_0-auc:0.92446\n",
      "[78]\tvalidation_0-auc:0.92434\n",
      "[79]\tvalidation_0-auc:0.92442\n",
      "[80]\tvalidation_0-auc:0.92459\n",
      "[81]\tvalidation_0-auc:0.92480\n",
      "[82]\tvalidation_0-auc:0.92504\n",
      "[83]\tvalidation_0-auc:0.92510\n",
      "[84]\tvalidation_0-auc:0.92528\n",
      "[85]\tvalidation_0-auc:0.92549\n",
      "[86]\tvalidation_0-auc:0.92556\n",
      "[87]\tvalidation_0-auc:0.92563\n",
      "[88]\tvalidation_0-auc:0.92562\n",
      "[89]\tvalidation_0-auc:0.92591\n",
      "[90]\tvalidation_0-auc:0.92595\n",
      "[91]\tvalidation_0-auc:0.92678\n",
      "[92]\tvalidation_0-auc:0.92699\n",
      "[93]\tvalidation_0-auc:0.92703\n",
      "[94]\tvalidation_0-auc:0.92713\n",
      "[95]\tvalidation_0-auc:0.92728\n",
      "[96]\tvalidation_0-auc:0.92721\n",
      "[97]\tvalidation_0-auc:0.92727\n",
      "[98]\tvalidation_0-auc:0.92715\n",
      "[99]\tvalidation_0-auc:0.92752\n",
      "[100]\tvalidation_0-auc:0.92775\n",
      "[101]\tvalidation_0-auc:0.92806\n",
      "[102]\tvalidation_0-auc:0.92821\n",
      "[103]\tvalidation_0-auc:0.92834\n",
      "[104]\tvalidation_0-auc:0.92830\n",
      "[105]\tvalidation_0-auc:0.92846\n",
      "[106]\tvalidation_0-auc:0.92855\n",
      "[107]\tvalidation_0-auc:0.92872\n",
      "[108]\tvalidation_0-auc:0.92886\n",
      "[109]\tvalidation_0-auc:0.92896\n",
      "[110]\tvalidation_0-auc:0.92941\n",
      "[111]\tvalidation_0-auc:0.92950\n",
      "[112]\tvalidation_0-auc:0.92944\n",
      "[113]\tvalidation_0-auc:0.92963\n",
      "[114]\tvalidation_0-auc:0.92963\n",
      "[115]\tvalidation_0-auc:0.92972\n",
      "[116]\tvalidation_0-auc:0.92971\n",
      "[117]\tvalidation_0-auc:0.92982\n",
      "[118]\tvalidation_0-auc:0.92988\n",
      "[119]\tvalidation_0-auc:0.92983\n",
      "[120]\tvalidation_0-auc:0.93019\n",
      "[121]\tvalidation_0-auc:0.93027\n",
      "[122]\tvalidation_0-auc:0.93058\n",
      "[123]\tvalidation_0-auc:0.93059\n",
      "[124]\tvalidation_0-auc:0.93052\n",
      "[125]\tvalidation_0-auc:0.93060\n",
      "[126]\tvalidation_0-auc:0.93055\n",
      "[127]\tvalidation_0-auc:0.93081\n",
      "[128]\tvalidation_0-auc:0.93089\n",
      "[129]\tvalidation_0-auc:0.93099\n",
      "[130]\tvalidation_0-auc:0.93096\n",
      "[131]\tvalidation_0-auc:0.93105\n",
      "[132]\tvalidation_0-auc:0.93124\n",
      "[133]\tvalidation_0-auc:0.93121\n",
      "[134]\tvalidation_0-auc:0.93125\n",
      "[135]\tvalidation_0-auc:0.93130\n",
      "[136]\tvalidation_0-auc:0.93131\n",
      "[137]\tvalidation_0-auc:0.93133\n",
      "[138]\tvalidation_0-auc:0.93129\n",
      "[139]\tvalidation_0-auc:0.93129\n",
      "[140]\tvalidation_0-auc:0.93138\n",
      "[141]\tvalidation_0-auc:0.93134\n",
      "[142]\tvalidation_0-auc:0.93139\n",
      "[143]\tvalidation_0-auc:0.93147\n",
      "[144]\tvalidation_0-auc:0.93152\n",
      "[145]\tvalidation_0-auc:0.93159\n",
      "[146]\tvalidation_0-auc:0.93152\n",
      "[147]\tvalidation_0-auc:0.93161\n",
      "[148]\tvalidation_0-auc:0.93160\n",
      "[149]\tvalidation_0-auc:0.93172\n",
      "[150]\tvalidation_0-auc:0.93173\n",
      "[151]\tvalidation_0-auc:0.93187\n",
      "[152]\tvalidation_0-auc:0.93191\n",
      "[153]\tvalidation_0-auc:0.93197\n",
      "[154]\tvalidation_0-auc:0.93198\n",
      "[155]\tvalidation_0-auc:0.93211\n",
      "[156]\tvalidation_0-auc:0.93218\n",
      "[157]\tvalidation_0-auc:0.93223\n",
      "[158]\tvalidation_0-auc:0.93223\n",
      "[159]\tvalidation_0-auc:0.93216\n",
      "[160]\tvalidation_0-auc:0.93235\n",
      "[161]\tvalidation_0-auc:0.93244\n",
      "[162]\tvalidation_0-auc:0.93242\n",
      "[163]\tvalidation_0-auc:0.93242\n",
      "[164]\tvalidation_0-auc:0.93247\n",
      "[165]\tvalidation_0-auc:0.93248\n",
      "[166]\tvalidation_0-auc:0.93243\n",
      "[167]\tvalidation_0-auc:0.93244\n",
      "[168]\tvalidation_0-auc:0.93248\n",
      "[169]\tvalidation_0-auc:0.93252\n",
      "[170]\tvalidation_0-auc:0.93253\n",
      "[171]\tvalidation_0-auc:0.93253\n",
      "[172]\tvalidation_0-auc:0.93250\n",
      "[173]\tvalidation_0-auc:0.93260\n",
      "[174]\tvalidation_0-auc:0.93268\n",
      "[175]\tvalidation_0-auc:0.93270\n",
      "[176]\tvalidation_0-auc:0.93270\n",
      "[177]\tvalidation_0-auc:0.93272\n",
      "[178]\tvalidation_0-auc:0.93271\n",
      "[179]\tvalidation_0-auc:0.93268\n",
      "[180]\tvalidation_0-auc:0.93271\n",
      "[181]\tvalidation_0-auc:0.93266\n",
      "[182]\tvalidation_0-auc:0.93260\n",
      "[183]\tvalidation_0-auc:0.93255\n",
      "[184]\tvalidation_0-auc:0.93256\n",
      "[185]\tvalidation_0-auc:0.93258\n",
      "[186]\tvalidation_0-auc:0.93263\n",
      "[187]\tvalidation_0-auc:0.93264\n",
      "[188]\tvalidation_0-auc:0.93259\n",
      "[189]\tvalidation_0-auc:0.93258\n",
      "[190]\tvalidation_0-auc:0.93263\n",
      "[191]\tvalidation_0-auc:0.93260\n",
      "[192]\tvalidation_0-auc:0.93260\n",
      "[193]\tvalidation_0-auc:0.93264\n",
      "[194]\tvalidation_0-auc:0.93264\n",
      "[195]\tvalidation_0-auc:0.93278\n",
      "[196]\tvalidation_0-auc:0.93281\n",
      "[197]\tvalidation_0-auc:0.93280\n",
      "[198]\tvalidation_0-auc:0.93278\n",
      "[199]\tvalidation_0-auc:0.93279\n",
      "[200]\tvalidation_0-auc:0.93279\n",
      "[201]\tvalidation_0-auc:0.93279\n",
      "[202]\tvalidation_0-auc:0.93281\n",
      "[203]\tvalidation_0-auc:0.93280\n",
      "[204]\tvalidation_0-auc:0.93282\n",
      "[205]\tvalidation_0-auc:0.93283\n",
      "[206]\tvalidation_0-auc:0.93283\n",
      "[207]\tvalidation_0-auc:0.93290\n",
      "[208]\tvalidation_0-auc:0.93286\n",
      "[209]\tvalidation_0-auc:0.93283\n",
      "[210]\tvalidation_0-auc:0.93285\n",
      "[211]\tvalidation_0-auc:0.93285\n",
      "[212]\tvalidation_0-auc:0.93286\n",
      "[213]\tvalidation_0-auc:0.93284\n",
      "[214]\tvalidation_0-auc:0.93281\n",
      "[215]\tvalidation_0-auc:0.93282\n",
      "[216]\tvalidation_0-auc:0.93281\n",
      "[217]\tvalidation_0-auc:0.93280\n",
      "[218]\tvalidation_0-auc:0.93279\n",
      "[219]\tvalidation_0-auc:0.93290\n",
      "[220]\tvalidation_0-auc:0.93287\n",
      "[221]\tvalidation_0-auc:0.93290\n",
      "[222]\tvalidation_0-auc:0.93290\n",
      "[223]\tvalidation_0-auc:0.93290\n",
      "[224]\tvalidation_0-auc:0.93291\n",
      "[225]\tvalidation_0-auc:0.93294\n",
      "[226]\tvalidation_0-auc:0.93292\n",
      "[227]\tvalidation_0-auc:0.93308\n",
      "[228]\tvalidation_0-auc:0.93307\n",
      "[229]\tvalidation_0-auc:0.93308\n",
      "[230]\tvalidation_0-auc:0.93307\n",
      "[231]\tvalidation_0-auc:0.93305\n",
      "[232]\tvalidation_0-auc:0.93306\n",
      "[233]\tvalidation_0-auc:0.93309\n",
      "[234]\tvalidation_0-auc:0.93308\n",
      "[235]\tvalidation_0-auc:0.93307\n",
      "[236]\tvalidation_0-auc:0.93320\n",
      "[237]\tvalidation_0-auc:0.93321\n",
      "[238]\tvalidation_0-auc:0.93325\n",
      "[239]\tvalidation_0-auc:0.93323\n",
      "[240]\tvalidation_0-auc:0.93329\n",
      "[241]\tvalidation_0-auc:0.93328\n",
      "[242]\tvalidation_0-auc:0.93327\n",
      "[243]\tvalidation_0-auc:0.93329\n",
      "[244]\tvalidation_0-auc:0.93329\n",
      "[245]\tvalidation_0-auc:0.93336\n",
      "[246]\tvalidation_0-auc:0.93343\n",
      "[247]\tvalidation_0-auc:0.93343\n",
      "[248]\tvalidation_0-auc:0.93341\n",
      "[249]\tvalidation_0-auc:0.93339\n",
      "[250]\tvalidation_0-auc:0.93342\n",
      "[251]\tvalidation_0-auc:0.93345\n",
      "[252]\tvalidation_0-auc:0.93346\n",
      "[253]\tvalidation_0-auc:0.93344\n",
      "[254]\tvalidation_0-auc:0.93345\n",
      "[255]\tvalidation_0-auc:0.93342\n",
      "[256]\tvalidation_0-auc:0.93345\n",
      "[257]\tvalidation_0-auc:0.93345\n",
      "[258]\tvalidation_0-auc:0.93347\n",
      "[259]\tvalidation_0-auc:0.93347\n",
      "[260]\tvalidation_0-auc:0.93348\n",
      "[261]\tvalidation_0-auc:0.93348\n",
      "[262]\tvalidation_0-auc:0.93347\n",
      "[263]\tvalidation_0-auc:0.93347\n",
      "[264]\tvalidation_0-auc:0.93346\n",
      "[265]\tvalidation_0-auc:0.93346\n",
      "[266]\tvalidation_0-auc:0.93344\n",
      "[267]\tvalidation_0-auc:0.93343\n",
      "[268]\tvalidation_0-auc:0.93351\n",
      "[269]\tvalidation_0-auc:0.93351\n",
      "[270]\tvalidation_0-auc:0.93353\n",
      "[271]\tvalidation_0-auc:0.93356\n",
      "[272]\tvalidation_0-auc:0.93358\n",
      "[273]\tvalidation_0-auc:0.93362\n",
      "[274]\tvalidation_0-auc:0.93368\n",
      "[275]\tvalidation_0-auc:0.93367\n",
      "[276]\tvalidation_0-auc:0.93366\n",
      "[277]\tvalidation_0-auc:0.93366\n",
      "[278]\tvalidation_0-auc:0.93376\n",
      "[279]\tvalidation_0-auc:0.93387\n",
      "[280]\tvalidation_0-auc:0.93389\n",
      "[281]\tvalidation_0-auc:0.93390\n",
      "[282]\tvalidation_0-auc:0.93388\n",
      "[283]\tvalidation_0-auc:0.93387\n",
      "[284]\tvalidation_0-auc:0.93387\n",
      "[285]\tvalidation_0-auc:0.93389\n",
      "[286]\tvalidation_0-auc:0.93398\n",
      "[287]\tvalidation_0-auc:0.93398\n",
      "[288]\tvalidation_0-auc:0.93396\n",
      "[289]\tvalidation_0-auc:0.93396\n",
      "[290]\tvalidation_0-auc:0.93397\n",
      "[291]\tvalidation_0-auc:0.93396\n",
      "[292]\tvalidation_0-auc:0.93397\n",
      "[293]\tvalidation_0-auc:0.93396\n",
      "[294]\tvalidation_0-auc:0.93393\n",
      "[295]\tvalidation_0-auc:0.93392\n",
      "[296]\tvalidation_0-auc:0.93392\n",
      "[297]\tvalidation_0-auc:0.93392\n",
      "[298]\tvalidation_0-auc:0.93392\n",
      "[299]\tvalidation_0-auc:0.93391\n",
      "[300]\tvalidation_0-auc:0.93397\n",
      "[301]\tvalidation_0-auc:0.93398\n",
      "[302]\tvalidation_0-auc:0.93402\n",
      "[303]\tvalidation_0-auc:0.93400\n",
      "[304]\tvalidation_0-auc:0.93400\n",
      "[305]\tvalidation_0-auc:0.93401\n",
      "[306]\tvalidation_0-auc:0.93404\n",
      "[307]\tvalidation_0-auc:0.93412\n",
      "[308]\tvalidation_0-auc:0.93413\n",
      "[309]\tvalidation_0-auc:0.93412\n",
      "[310]\tvalidation_0-auc:0.93415\n",
      "[311]\tvalidation_0-auc:0.93414\n",
      "[312]\tvalidation_0-auc:0.93416\n",
      "[313]\tvalidation_0-auc:0.93416\n",
      "[314]\tvalidation_0-auc:0.93416\n",
      "[315]\tvalidation_0-auc:0.93420\n",
      "[316]\tvalidation_0-auc:0.93416\n",
      "[317]\tvalidation_0-auc:0.93417\n",
      "[318]\tvalidation_0-auc:0.93424\n",
      "[319]\tvalidation_0-auc:0.93424\n",
      "[320]\tvalidation_0-auc:0.93429\n",
      "[321]\tvalidation_0-auc:0.93429\n",
      "[322]\tvalidation_0-auc:0.93432\n",
      "[323]\tvalidation_0-auc:0.93430\n",
      "[324]\tvalidation_0-auc:0.93431\n",
      "[325]\tvalidation_0-auc:0.93433\n",
      "[326]\tvalidation_0-auc:0.93432\n",
      "[327]\tvalidation_0-auc:0.93431\n",
      "[328]\tvalidation_0-auc:0.93434\n",
      "[329]\tvalidation_0-auc:0.93436\n",
      "[330]\tvalidation_0-auc:0.93434\n",
      "[331]\tvalidation_0-auc:0.93432\n",
      "[332]\tvalidation_0-auc:0.93432\n",
      "[333]\tvalidation_0-auc:0.93433\n",
      "[334]\tvalidation_0-auc:0.93429\n",
      "[335]\tvalidation_0-auc:0.93431\n",
      "[336]\tvalidation_0-auc:0.93432\n",
      "[337]\tvalidation_0-auc:0.93433\n",
      "[338]\tvalidation_0-auc:0.93435\n",
      "[339]\tvalidation_0-auc:0.93435\n",
      "[340]\tvalidation_0-auc:0.93435\n",
      "[341]\tvalidation_0-auc:0.93435\n",
      "[342]\tvalidation_0-auc:0.93436\n",
      "[343]\tvalidation_0-auc:0.93435\n",
      "[344]\tvalidation_0-auc:0.93433\n",
      "[345]\tvalidation_0-auc:0.93432\n",
      "[346]\tvalidation_0-auc:0.93434\n",
      "[347]\tvalidation_0-auc:0.93434\n",
      "[348]\tvalidation_0-auc:0.93433\n",
      "[349]\tvalidation_0-auc:0.93431\n",
      "[350]\tvalidation_0-auc:0.93432\n",
      "[351]\tvalidation_0-auc:0.93431\n",
      "[352]\tvalidation_0-auc:0.93431\n",
      "[353]\tvalidation_0-auc:0.93430\n",
      "[354]\tvalidation_0-auc:0.93431\n",
      "[355]\tvalidation_0-auc:0.93431\n",
      "[356]\tvalidation_0-auc:0.93428\n",
      "[357]\tvalidation_0-auc:0.93425\n",
      "[358]\tvalidation_0-auc:0.93424\n",
      "[359]\tvalidation_0-auc:0.93426\n",
      "[360]\tvalidation_0-auc:0.93424\n",
      "[361]\tvalidation_0-auc:0.93423\n",
      "[362]\tvalidation_0-auc:0.93422\n",
      "[363]\tvalidation_0-auc:0.93424\n",
      "[364]\tvalidation_0-auc:0.93431\n",
      "[365]\tvalidation_0-auc:0.93432\n",
      "[366]\tvalidation_0-auc:0.93431\n",
      "[367]\tvalidation_0-auc:0.93431\n",
      "[368]\tvalidation_0-auc:0.93430\n",
      "[369]\tvalidation_0-auc:0.93429\n",
      "[370]\tvalidation_0-auc:0.93431\n",
      "[371]\tvalidation_0-auc:0.93433\n",
      "[372]\tvalidation_0-auc:0.93432\n",
      "[373]\tvalidation_0-auc:0.93435\n",
      "[374]\tvalidation_0-auc:0.93436\n",
      "[375]\tvalidation_0-auc:0.93437\n",
      "[376]\tvalidation_0-auc:0.93437\n",
      "[377]\tvalidation_0-auc:0.93437\n",
      "[378]\tvalidation_0-auc:0.93434\n",
      "[379]\tvalidation_0-auc:0.93435\n",
      "[380]\tvalidation_0-auc:0.93435\n",
      "[381]\tvalidation_0-auc:0.93439\n",
      "[382]\tvalidation_0-auc:0.93443\n",
      "[383]\tvalidation_0-auc:0.93443\n",
      "[384]\tvalidation_0-auc:0.93443\n",
      "[385]\tvalidation_0-auc:0.93449\n",
      "[386]\tvalidation_0-auc:0.93456\n",
      "[387]\tvalidation_0-auc:0.93456\n",
      "[388]\tvalidation_0-auc:0.93450\n",
      "[389]\tvalidation_0-auc:0.93450\n",
      "[390]\tvalidation_0-auc:0.93450\n",
      "[391]\tvalidation_0-auc:0.93452\n",
      "[392]\tvalidation_0-auc:0.93456\n",
      "[393]\tvalidation_0-auc:0.93461\n",
      "[394]\tvalidation_0-auc:0.93463\n",
      "[395]\tvalidation_0-auc:0.93463\n",
      "[396]\tvalidation_0-auc:0.93467\n",
      "[397]\tvalidation_0-auc:0.93465\n",
      "[398]\tvalidation_0-auc:0.93464\n",
      "[399]\tvalidation_0-auc:0.93466\n",
      "[400]\tvalidation_0-auc:0.93467\n",
      "[401]\tvalidation_0-auc:0.93468\n",
      "[402]\tvalidation_0-auc:0.93469\n",
      "[403]\tvalidation_0-auc:0.93467\n",
      "[404]\tvalidation_0-auc:0.93464\n",
      "[405]\tvalidation_0-auc:0.93464\n",
      "[406]\tvalidation_0-auc:0.93463\n",
      "[407]\tvalidation_0-auc:0.93463\n",
      "[408]\tvalidation_0-auc:0.93468\n",
      "[409]\tvalidation_0-auc:0.93467\n",
      "[410]\tvalidation_0-auc:0.93465\n",
      "[411]\tvalidation_0-auc:0.93464\n",
      "[412]\tvalidation_0-auc:0.93470\n",
      "[413]\tvalidation_0-auc:0.93473\n",
      "[414]\tvalidation_0-auc:0.93472\n",
      "[415]\tvalidation_0-auc:0.93471\n",
      "[416]\tvalidation_0-auc:0.93476\n",
      "[417]\tvalidation_0-auc:0.93475\n",
      "[418]\tvalidation_0-auc:0.93472\n",
      "[419]\tvalidation_0-auc:0.93472\n",
      "[420]\tvalidation_0-auc:0.93468\n",
      "[421]\tvalidation_0-auc:0.93469\n",
      "[422]\tvalidation_0-auc:0.93476\n",
      "[423]\tvalidation_0-auc:0.93475\n",
      "[424]\tvalidation_0-auc:0.93477\n",
      "[425]\tvalidation_0-auc:0.93479\n",
      "[426]\tvalidation_0-auc:0.93476\n",
      "[427]\tvalidation_0-auc:0.93477\n",
      "[428]\tvalidation_0-auc:0.93477\n",
      "[429]\tvalidation_0-auc:0.93479\n",
      "[430]\tvalidation_0-auc:0.93480\n",
      "[431]\tvalidation_0-auc:0.93485\n",
      "[432]\tvalidation_0-auc:0.93484\n",
      "[433]\tvalidation_0-auc:0.93487\n",
      "[434]\tvalidation_0-auc:0.93487\n",
      "[435]\tvalidation_0-auc:0.93490\n",
      "[436]\tvalidation_0-auc:0.93489\n",
      "[437]\tvalidation_0-auc:0.93489\n",
      "[438]\tvalidation_0-auc:0.93489\n",
      "[439]\tvalidation_0-auc:0.93485\n",
      "[440]\tvalidation_0-auc:0.93485\n",
      "[441]\tvalidation_0-auc:0.93488\n",
      "[442]\tvalidation_0-auc:0.93486\n",
      "[443]\tvalidation_0-auc:0.93484\n",
      "[444]\tvalidation_0-auc:0.93487\n",
      "[445]\tvalidation_0-auc:0.93488\n",
      "[446]\tvalidation_0-auc:0.93485\n",
      "[447]\tvalidation_0-auc:0.93488\n",
      "[448]\tvalidation_0-auc:0.93490\n",
      "[449]\tvalidation_0-auc:0.93493\n",
      "[450]\tvalidation_0-auc:0.93495\n",
      "[451]\tvalidation_0-auc:0.93494\n",
      "[452]\tvalidation_0-auc:0.93495\n",
      "[453]\tvalidation_0-auc:0.93499\n",
      "[454]\tvalidation_0-auc:0.93498\n",
      "[455]\tvalidation_0-auc:0.93497\n",
      "[456]\tvalidation_0-auc:0.93500\n",
      "[457]\tvalidation_0-auc:0.93501\n",
      "[458]\tvalidation_0-auc:0.93500\n",
      "[459]\tvalidation_0-auc:0.93501\n",
      "[460]\tvalidation_0-auc:0.93500\n",
      "[461]\tvalidation_0-auc:0.93502\n",
      "[462]\tvalidation_0-auc:0.93504\n",
      "[463]\tvalidation_0-auc:0.93504\n",
      "[464]\tvalidation_0-auc:0.93502\n",
      "[465]\tvalidation_0-auc:0.93502\n",
      "[466]\tvalidation_0-auc:0.93501\n",
      "[467]\tvalidation_0-auc:0.93505\n",
      "[468]\tvalidation_0-auc:0.93505\n",
      "[469]\tvalidation_0-auc:0.93505\n",
      "[470]\tvalidation_0-auc:0.93503\n",
      "[471]\tvalidation_0-auc:0.93504\n",
      "[472]\tvalidation_0-auc:0.93504\n",
      "[473]\tvalidation_0-auc:0.93506\n",
      "[474]\tvalidation_0-auc:0.93504\n",
      "[475]\tvalidation_0-auc:0.93505\n",
      "[476]\tvalidation_0-auc:0.93503\n",
      "[477]\tvalidation_0-auc:0.93503\n",
      "[478]\tvalidation_0-auc:0.93502\n",
      "[479]\tvalidation_0-auc:0.93501\n",
      "[480]\tvalidation_0-auc:0.93500\n",
      "[481]\tvalidation_0-auc:0.93499\n",
      "[482]\tvalidation_0-auc:0.93500\n",
      "[483]\tvalidation_0-auc:0.93499\n",
      "[484]\tvalidation_0-auc:0.93498\n",
      "[485]\tvalidation_0-auc:0.93498\n",
      "[486]\tvalidation_0-auc:0.93497\n",
      "[487]\tvalidation_0-auc:0.93497\n",
      "[488]\tvalidation_0-auc:0.93498\n",
      "[489]\tvalidation_0-auc:0.93494\n",
      "[490]\tvalidation_0-auc:0.93493\n",
      "[491]\tvalidation_0-auc:0.93492\n",
      "[492]\tvalidation_0-auc:0.93495\n",
      "[493]\tvalidation_0-auc:0.93495\n",
      "[494]\tvalidation_0-auc:0.93492\n",
      "[495]\tvalidation_0-auc:0.93492\n",
      "[496]\tvalidation_0-auc:0.93490\n",
      "[497]\tvalidation_0-auc:0.93495\n",
      "[498]\tvalidation_0-auc:0.93494\n",
      "[499]\tvalidation_0-auc:0.93494\n",
      "[500]\tvalidation_0-auc:0.93494\n",
      "[501]\tvalidation_0-auc:0.93494\n",
      "[502]\tvalidation_0-auc:0.93494\n",
      "[503]\tvalidation_0-auc:0.93494\n",
      "[504]\tvalidation_0-auc:0.93497\n",
      "[505]\tvalidation_0-auc:0.93498\n",
      "[506]\tvalidation_0-auc:0.93498\n",
      "[507]\tvalidation_0-auc:0.93495\n",
      "[508]\tvalidation_0-auc:0.93496\n",
      "[509]\tvalidation_0-auc:0.93499\n",
      "[510]\tvalidation_0-auc:0.93499\n",
      "[511]\tvalidation_0-auc:0.93499\n",
      "[512]\tvalidation_0-auc:0.93498\n",
      "[513]\tvalidation_0-auc:0.93500\n",
      "[514]\tvalidation_0-auc:0.93499\n",
      "[515]\tvalidation_0-auc:0.93498\n",
      "[516]\tvalidation_0-auc:0.93498\n",
      "[517]\tvalidation_0-auc:0.93501\n",
      "[518]\tvalidation_0-auc:0.93500\n",
      "[519]\tvalidation_0-auc:0.93501\n",
      "[520]\tvalidation_0-auc:0.93503\n",
      "[521]\tvalidation_0-auc:0.93503\n",
      "[522]\tvalidation_0-auc:0.93506\n",
      "[523]\tvalidation_0-auc:0.93506\n",
      "[524]\tvalidation_0-auc:0.93506\n",
      "[525]\tvalidation_0-auc:0.93508\n",
      "[526]\tvalidation_0-auc:0.93506\n",
      "[527]\tvalidation_0-auc:0.93506\n",
      "[528]\tvalidation_0-auc:0.93509\n",
      "[529]\tvalidation_0-auc:0.93510\n",
      "[530]\tvalidation_0-auc:0.93510\n",
      "[531]\tvalidation_0-auc:0.93512\n",
      "[532]\tvalidation_0-auc:0.93510\n",
      "[533]\tvalidation_0-auc:0.93506\n",
      "[534]\tvalidation_0-auc:0.93505\n",
      "[535]\tvalidation_0-auc:0.93504\n",
      "[536]\tvalidation_0-auc:0.93508\n",
      "[537]\tvalidation_0-auc:0.93508\n",
      "[538]\tvalidation_0-auc:0.93508\n",
      "[539]\tvalidation_0-auc:0.93508\n",
      "[540]\tvalidation_0-auc:0.93510\n",
      "[541]\tvalidation_0-auc:0.93507\n",
      "[542]\tvalidation_0-auc:0.93508\n",
      "[543]\tvalidation_0-auc:0.93509\n",
      "[544]\tvalidation_0-auc:0.93507\n",
      "[545]\tvalidation_0-auc:0.93507\n",
      "[546]\tvalidation_0-auc:0.93506\n",
      "[547]\tvalidation_0-auc:0.93505\n",
      "[548]\tvalidation_0-auc:0.93506\n",
      "[549]\tvalidation_0-auc:0.93506\n",
      "[550]\tvalidation_0-auc:0.93506\n",
      "[551]\tvalidation_0-auc:0.93507\n",
      "[552]\tvalidation_0-auc:0.93507\n",
      "[553]\tvalidation_0-auc:0.93507\n",
      "[554]\tvalidation_0-auc:0.93508\n",
      "[555]\tvalidation_0-auc:0.93508\n",
      "[556]\tvalidation_0-auc:0.93508\n",
      "[557]\tvalidation_0-auc:0.93506\n",
      "[558]\tvalidation_0-auc:0.93508\n",
      "[559]\tvalidation_0-auc:0.93507\n",
      "[560]\tvalidation_0-auc:0.93506\n",
      "[561]\tvalidation_0-auc:0.93507\n",
      "[562]\tvalidation_0-auc:0.93507\n",
      "[563]\tvalidation_0-auc:0.93506\n",
      "[564]\tvalidation_0-auc:0.93506\n",
      "[565]\tvalidation_0-auc:0.93506\n",
      "[566]\tvalidation_0-auc:0.93508\n",
      "[567]\tvalidation_0-auc:0.93506\n",
      "[568]\tvalidation_0-auc:0.93507\n",
      "[569]\tvalidation_0-auc:0.93507\n",
      "[570]\tvalidation_0-auc:0.93507\n",
      "[571]\tvalidation_0-auc:0.93506\n",
      "[572]\tvalidation_0-auc:0.93505\n",
      "[573]\tvalidation_0-auc:0.93506\n",
      "[574]\tvalidation_0-auc:0.93505\n",
      "[575]\tvalidation_0-auc:0.93508\n",
      "[576]\tvalidation_0-auc:0.93508\n",
      "[577]\tvalidation_0-auc:0.93509\n",
      "[578]\tvalidation_0-auc:0.93509\n",
      "[579]\tvalidation_0-auc:0.93508\n",
      "[580]\tvalidation_0-auc:0.93509\n",
      "[581]\tvalidation_0-auc:0.93510\n",
      "[582]\tvalidation_0-auc:0.93509\n",
      "[583]\tvalidation_0-auc:0.93509\n",
      "[584]\tvalidation_0-auc:0.93510\n",
      "[585]\tvalidation_0-auc:0.93510\n",
      "[586]\tvalidation_0-auc:0.93509\n",
      "[587]\tvalidation_0-auc:0.93509\n",
      "[588]\tvalidation_0-auc:0.93510\n",
      "[589]\tvalidation_0-auc:0.93509\n",
      "[590]\tvalidation_0-auc:0.93509\n",
      "[591]\tvalidation_0-auc:0.93509\n",
      "[592]\tvalidation_0-auc:0.93515\n",
      "[593]\tvalidation_0-auc:0.93514\n",
      "[594]\tvalidation_0-auc:0.93514\n",
      "[595]\tvalidation_0-auc:0.93514\n",
      "[596]\tvalidation_0-auc:0.93514\n",
      "[597]\tvalidation_0-auc:0.93513\n",
      "[598]\tvalidation_0-auc:0.93512\n",
      "[599]\tvalidation_0-auc:0.93512\n"
     ]
    }
   ],
   "source": [
    "# Hyperparameter tuning and cross validation\n",
    "tune_model = grid_search.fit(X = x2_train,y = y2_train,**evaluation_parameters)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "id": "prerequisite-helping",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'booster': 'gbtree', 'colsample_bytree': 0.5, 'gamma': 0, 'learning_rate': 0.05, 'max_depth': 6, 'min_child_weight': 1, 'n_estimators': 600, 'random_state': 1502, 'subsample': 1} 0.9361644248601657\n"
     ]
    }
   ],
   "source": [
    "print(grid_search.best_params_, grid_search.best_score_ )"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fresh-rubber",
   "metadata": {},
   "source": [
    "### Try second cross validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "afraid-analyst",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Setting the cross validation parameters\n",
    "from sklearn.model_selection import KFold\n",
    "tune_control = KFold(n_splits = 5,\n",
    "                     shuffle = True,\n",
    "                     random_state = 1502).split(X = x2_train,\n",
    "                                                y = y2_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "liquid-genealogy",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set parameter tuning part2\n",
    "\n",
    "tune_grid2 = {'learning_rate': [0.05],\n",
    "               'max_depth': [6],\n",
    "               'colsample_bytree': [0.5],\n",
    "               'subsample': [0.9, 1],\n",
    "               'min_child_weight': range(1,5,1),\n",
    "               'gamma': [0, 0.1], \n",
    "               'random_state': [1502],\n",
    "               'n_estimators': range(200, 2000, 200),\n",
    "               'booster': [\"gbtree\"]}                                                "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "filled-laptop",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Cross Validation Assembly\n",
    "\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "grid_search2 = GridSearchCV(estimator = classifier,\n",
    "                           param_grid = tune_grid2,\n",
    "                            scoring = \"roc_auc\",\n",
    "                            n_jobs = 6,\n",
    "                            cv = tune_control,\n",
    "                            verbose = 5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "applied-large",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 144 candidates, totalling 720 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=6)]: Using backend LokyBackend with 6 concurrent workers.\n",
      "[Parallel(n_jobs=6)]: Done   6 tasks      | elapsed:   30.3s\n",
      "[Parallel(n_jobs=6)]: Done  60 tasks      | elapsed:  9.4min\n",
      "[Parallel(n_jobs=6)]: Done 150 tasks      | elapsed: 24.3min\n",
      "[Parallel(n_jobs=6)]: Done 276 tasks      | elapsed: 43.4min\n",
      "[Parallel(n_jobs=6)]: Done 438 tasks      | elapsed: 67.1min\n",
      "[Parallel(n_jobs=6)]: Done 636 tasks      | elapsed: 98.6min\n",
      "[Parallel(n_jobs=6)]: Done 720 out of 720 | elapsed: 112.9min finished\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0]\tvalidation_0-auc:0.85589\n",
      "Will train until validation_0-auc hasn't improved in 100 rounds.\n",
      "[1]\tvalidation_0-auc:0.86457\n",
      "[2]\tvalidation_0-auc:0.89279\n",
      "[3]\tvalidation_0-auc:0.89654\n",
      "[4]\tvalidation_0-auc:0.90140\n",
      "[5]\tvalidation_0-auc:0.90268\n",
      "[6]\tvalidation_0-auc:0.90453\n",
      "[7]\tvalidation_0-auc:0.90749\n",
      "[8]\tvalidation_0-auc:0.90902\n",
      "[9]\tvalidation_0-auc:0.90909\n",
      "[10]\tvalidation_0-auc:0.90984\n",
      "[11]\tvalidation_0-auc:0.91039\n",
      "[12]\tvalidation_0-auc:0.91199\n",
      "[13]\tvalidation_0-auc:0.91362\n",
      "[14]\tvalidation_0-auc:0.91411\n",
      "[15]\tvalidation_0-auc:0.91410\n",
      "[16]\tvalidation_0-auc:0.91367\n",
      "[17]\tvalidation_0-auc:0.91389\n",
      "[18]\tvalidation_0-auc:0.91339\n",
      "[19]\tvalidation_0-auc:0.91431\n",
      "[20]\tvalidation_0-auc:0.91440\n",
      "[21]\tvalidation_0-auc:0.91465\n",
      "[22]\tvalidation_0-auc:0.91507\n",
      "[23]\tvalidation_0-auc:0.91604\n",
      "[24]\tvalidation_0-auc:0.91624\n",
      "[25]\tvalidation_0-auc:0.91696\n",
      "[26]\tvalidation_0-auc:0.91738\n",
      "[27]\tvalidation_0-auc:0.91732\n",
      "[28]\tvalidation_0-auc:0.91745\n",
      "[29]\tvalidation_0-auc:0.91829\n",
      "[30]\tvalidation_0-auc:0.91814\n",
      "[31]\tvalidation_0-auc:0.91843\n",
      "[32]\tvalidation_0-auc:0.91854\n",
      "[33]\tvalidation_0-auc:0.91916\n",
      "[34]\tvalidation_0-auc:0.91924\n",
      "[35]\tvalidation_0-auc:0.91973\n",
      "[36]\tvalidation_0-auc:0.91992\n",
      "[37]\tvalidation_0-auc:0.92026\n",
      "[38]\tvalidation_0-auc:0.92054\n",
      "[39]\tvalidation_0-auc:0.92084\n",
      "[40]\tvalidation_0-auc:0.92161\n",
      "[41]\tvalidation_0-auc:0.92180\n",
      "[42]\tvalidation_0-auc:0.92192\n",
      "[43]\tvalidation_0-auc:0.92184\n",
      "[44]\tvalidation_0-auc:0.92190\n",
      "[45]\tvalidation_0-auc:0.92219\n",
      "[46]\tvalidation_0-auc:0.92219\n",
      "[47]\tvalidation_0-auc:0.92260\n",
      "[48]\tvalidation_0-auc:0.92271\n",
      "[49]\tvalidation_0-auc:0.92279\n",
      "[50]\tvalidation_0-auc:0.92302\n",
      "[51]\tvalidation_0-auc:0.92324\n",
      "[52]\tvalidation_0-auc:0.92348\n",
      "[53]\tvalidation_0-auc:0.92360\n",
      "[54]\tvalidation_0-auc:0.92358\n",
      "[55]\tvalidation_0-auc:0.92359\n",
      "[56]\tvalidation_0-auc:0.92360\n",
      "[57]\tvalidation_0-auc:0.92350\n",
      "[58]\tvalidation_0-auc:0.92352\n",
      "[59]\tvalidation_0-auc:0.92356\n",
      "[60]\tvalidation_0-auc:0.92347\n",
      "[61]\tvalidation_0-auc:0.92324\n",
      "[62]\tvalidation_0-auc:0.92357\n",
      "[63]\tvalidation_0-auc:0.92369\n",
      "[64]\tvalidation_0-auc:0.92362\n",
      "[65]\tvalidation_0-auc:0.92410\n",
      "[66]\tvalidation_0-auc:0.92403\n",
      "[67]\tvalidation_0-auc:0.92419\n",
      "[68]\tvalidation_0-auc:0.92431\n",
      "[69]\tvalidation_0-auc:0.92417\n",
      "[70]\tvalidation_0-auc:0.92398\n",
      "[71]\tvalidation_0-auc:0.92426\n",
      "[72]\tvalidation_0-auc:0.92435\n",
      "[73]\tvalidation_0-auc:0.92450\n",
      "[74]\tvalidation_0-auc:0.92459\n",
      "[75]\tvalidation_0-auc:0.92462\n",
      "[76]\tvalidation_0-auc:0.92481\n",
      "[77]\tvalidation_0-auc:0.92485\n",
      "[78]\tvalidation_0-auc:0.92488\n",
      "[79]\tvalidation_0-auc:0.92494\n",
      "[80]\tvalidation_0-auc:0.92504\n",
      "[81]\tvalidation_0-auc:0.92528\n",
      "[82]\tvalidation_0-auc:0.92549\n",
      "[83]\tvalidation_0-auc:0.92561\n",
      "[84]\tvalidation_0-auc:0.92579\n",
      "[85]\tvalidation_0-auc:0.92595\n",
      "[86]\tvalidation_0-auc:0.92594\n",
      "[87]\tvalidation_0-auc:0.92602\n",
      "[88]\tvalidation_0-auc:0.92602\n",
      "[89]\tvalidation_0-auc:0.92628\n",
      "[90]\tvalidation_0-auc:0.92641\n",
      "[91]\tvalidation_0-auc:0.92719\n",
      "[92]\tvalidation_0-auc:0.92737\n",
      "[93]\tvalidation_0-auc:0.92744\n",
      "[94]\tvalidation_0-auc:0.92754\n",
      "[95]\tvalidation_0-auc:0.92769\n",
      "[96]\tvalidation_0-auc:0.92755\n",
      "[97]\tvalidation_0-auc:0.92762\n",
      "[98]\tvalidation_0-auc:0.92756\n",
      "[99]\tvalidation_0-auc:0.92786\n",
      "[100]\tvalidation_0-auc:0.92812\n",
      "[101]\tvalidation_0-auc:0.92831\n",
      "[102]\tvalidation_0-auc:0.92848\n",
      "[103]\tvalidation_0-auc:0.92865\n",
      "[104]\tvalidation_0-auc:0.92878\n",
      "[105]\tvalidation_0-auc:0.92897\n",
      "[106]\tvalidation_0-auc:0.92915\n",
      "[107]\tvalidation_0-auc:0.92929\n",
      "[108]\tvalidation_0-auc:0.92941\n",
      "[109]\tvalidation_0-auc:0.92948\n",
      "[110]\tvalidation_0-auc:0.92993\n",
      "[111]\tvalidation_0-auc:0.93003\n",
      "[112]\tvalidation_0-auc:0.92994\n",
      "[113]\tvalidation_0-auc:0.93011\n",
      "[114]\tvalidation_0-auc:0.93007\n",
      "[115]\tvalidation_0-auc:0.93012\n",
      "[116]\tvalidation_0-auc:0.93017\n",
      "[117]\tvalidation_0-auc:0.93031\n",
      "[118]\tvalidation_0-auc:0.93034\n",
      "[119]\tvalidation_0-auc:0.93026\n",
      "[120]\tvalidation_0-auc:0.93061\n",
      "[121]\tvalidation_0-auc:0.93068\n",
      "[122]\tvalidation_0-auc:0.93101\n",
      "[123]\tvalidation_0-auc:0.93103\n",
      "[124]\tvalidation_0-auc:0.93104\n",
      "[125]\tvalidation_0-auc:0.93110\n",
      "[126]\tvalidation_0-auc:0.93105\n",
      "[127]\tvalidation_0-auc:0.93129\n",
      "[128]\tvalidation_0-auc:0.93135\n",
      "[129]\tvalidation_0-auc:0.93144\n",
      "[130]\tvalidation_0-auc:0.93141\n",
      "[131]\tvalidation_0-auc:0.93144\n",
      "[132]\tvalidation_0-auc:0.93147\n",
      "[133]\tvalidation_0-auc:0.93146\n",
      "[134]\tvalidation_0-auc:0.93163\n",
      "[135]\tvalidation_0-auc:0.93165\n",
      "[136]\tvalidation_0-auc:0.93163\n",
      "[137]\tvalidation_0-auc:0.93167\n",
      "[138]\tvalidation_0-auc:0.93165\n",
      "[139]\tvalidation_0-auc:0.93166\n",
      "[140]\tvalidation_0-auc:0.93172\n",
      "[141]\tvalidation_0-auc:0.93167\n",
      "[142]\tvalidation_0-auc:0.93175\n",
      "[143]\tvalidation_0-auc:0.93189\n",
      "[144]\tvalidation_0-auc:0.93189\n",
      "[145]\tvalidation_0-auc:0.93191\n",
      "[146]\tvalidation_0-auc:0.93186\n",
      "[147]\tvalidation_0-auc:0.93182\n",
      "[148]\tvalidation_0-auc:0.93182\n",
      "[149]\tvalidation_0-auc:0.93187\n",
      "[150]\tvalidation_0-auc:0.93188\n",
      "[151]\tvalidation_0-auc:0.93197\n",
      "[152]\tvalidation_0-auc:0.93200\n",
      "[153]\tvalidation_0-auc:0.93206\n",
      "[154]\tvalidation_0-auc:0.93205\n",
      "[155]\tvalidation_0-auc:0.93217\n",
      "[156]\tvalidation_0-auc:0.93226\n",
      "[157]\tvalidation_0-auc:0.93248\n",
      "[158]\tvalidation_0-auc:0.93248\n",
      "[159]\tvalidation_0-auc:0.93249\n",
      "[160]\tvalidation_0-auc:0.93263\n",
      "[161]\tvalidation_0-auc:0.93273\n",
      "[162]\tvalidation_0-auc:0.93273\n",
      "[163]\tvalidation_0-auc:0.93272\n",
      "[164]\tvalidation_0-auc:0.93277\n",
      "[165]\tvalidation_0-auc:0.93280\n",
      "[166]\tvalidation_0-auc:0.93277\n",
      "[167]\tvalidation_0-auc:0.93275\n",
      "[168]\tvalidation_0-auc:0.93277\n",
      "[169]\tvalidation_0-auc:0.93275\n",
      "[170]\tvalidation_0-auc:0.93278\n",
      "[171]\tvalidation_0-auc:0.93283\n",
      "[172]\tvalidation_0-auc:0.93287\n",
      "[173]\tvalidation_0-auc:0.93284\n",
      "[174]\tvalidation_0-auc:0.93284\n",
      "[175]\tvalidation_0-auc:0.93287\n",
      "[176]\tvalidation_0-auc:0.93292\n",
      "[177]\tvalidation_0-auc:0.93293\n",
      "[178]\tvalidation_0-auc:0.93302\n",
      "[179]\tvalidation_0-auc:0.93298\n",
      "[180]\tvalidation_0-auc:0.93319\n",
      "[181]\tvalidation_0-auc:0.93316\n",
      "[182]\tvalidation_0-auc:0.93323\n",
      "[183]\tvalidation_0-auc:0.93327\n",
      "[184]\tvalidation_0-auc:0.93331\n",
      "[185]\tvalidation_0-auc:0.93336\n",
      "[186]\tvalidation_0-auc:0.93332\n",
      "[187]\tvalidation_0-auc:0.93332\n",
      "[188]\tvalidation_0-auc:0.93327\n",
      "[189]\tvalidation_0-auc:0.93332\n",
      "[190]\tvalidation_0-auc:0.93336\n",
      "[191]\tvalidation_0-auc:0.93331\n",
      "[192]\tvalidation_0-auc:0.93333\n",
      "[193]\tvalidation_0-auc:0.93333\n",
      "[194]\tvalidation_0-auc:0.93334\n",
      "[195]\tvalidation_0-auc:0.93335\n",
      "[196]\tvalidation_0-auc:0.93338\n",
      "[197]\tvalidation_0-auc:0.93336\n",
      "[198]\tvalidation_0-auc:0.93335\n",
      "[199]\tvalidation_0-auc:0.93339\n",
      "[200]\tvalidation_0-auc:0.93338\n",
      "[201]\tvalidation_0-auc:0.93338\n",
      "[202]\tvalidation_0-auc:0.93339\n",
      "[203]\tvalidation_0-auc:0.93337\n",
      "[204]\tvalidation_0-auc:0.93337\n",
      "[205]\tvalidation_0-auc:0.93337\n",
      "[206]\tvalidation_0-auc:0.93337\n",
      "[207]\tvalidation_0-auc:0.93339\n",
      "[208]\tvalidation_0-auc:0.93340\n",
      "[209]\tvalidation_0-auc:0.93340\n",
      "[210]\tvalidation_0-auc:0.93341\n",
      "[211]\tvalidation_0-auc:0.93344\n",
      "[212]\tvalidation_0-auc:0.93341\n",
      "[213]\tvalidation_0-auc:0.93339\n",
      "[214]\tvalidation_0-auc:0.93337\n",
      "[215]\tvalidation_0-auc:0.93338\n",
      "[216]\tvalidation_0-auc:0.93340\n",
      "[217]\tvalidation_0-auc:0.93340\n",
      "[218]\tvalidation_0-auc:0.93340\n",
      "[219]\tvalidation_0-auc:0.93341\n",
      "[220]\tvalidation_0-auc:0.93354\n",
      "[221]\tvalidation_0-auc:0.93355\n",
      "[222]\tvalidation_0-auc:0.93354\n",
      "[223]\tvalidation_0-auc:0.93353\n",
      "[224]\tvalidation_0-auc:0.93354\n",
      "[225]\tvalidation_0-auc:0.93358\n",
      "[226]\tvalidation_0-auc:0.93357\n",
      "[227]\tvalidation_0-auc:0.93355\n",
      "[228]\tvalidation_0-auc:0.93355\n",
      "[229]\tvalidation_0-auc:0.93356\n",
      "[230]\tvalidation_0-auc:0.93356\n",
      "[231]\tvalidation_0-auc:0.93354\n",
      "[232]\tvalidation_0-auc:0.93356\n",
      "[233]\tvalidation_0-auc:0.93361\n",
      "[234]\tvalidation_0-auc:0.93359\n",
      "[235]\tvalidation_0-auc:0.93358\n",
      "[236]\tvalidation_0-auc:0.93368\n",
      "[237]\tvalidation_0-auc:0.93378\n",
      "[238]\tvalidation_0-auc:0.93378\n",
      "[239]\tvalidation_0-auc:0.93381\n",
      "[240]\tvalidation_0-auc:0.93386\n",
      "[241]\tvalidation_0-auc:0.93385\n",
      "[242]\tvalidation_0-auc:0.93383\n",
      "[243]\tvalidation_0-auc:0.93383\n",
      "[244]\tvalidation_0-auc:0.93383\n",
      "[245]\tvalidation_0-auc:0.93385\n",
      "[246]\tvalidation_0-auc:0.93384\n",
      "[247]\tvalidation_0-auc:0.93383\n",
      "[248]\tvalidation_0-auc:0.93382\n",
      "[249]\tvalidation_0-auc:0.93381\n",
      "[250]\tvalidation_0-auc:0.93378\n",
      "[251]\tvalidation_0-auc:0.93382\n",
      "[252]\tvalidation_0-auc:0.93382\n",
      "[253]\tvalidation_0-auc:0.93385\n",
      "[254]\tvalidation_0-auc:0.93384\n",
      "[255]\tvalidation_0-auc:0.93385\n",
      "[256]\tvalidation_0-auc:0.93397\n",
      "[257]\tvalidation_0-auc:0.93398\n",
      "[258]\tvalidation_0-auc:0.93398\n",
      "[259]\tvalidation_0-auc:0.93395\n",
      "[260]\tvalidation_0-auc:0.93397\n",
      "[261]\tvalidation_0-auc:0.93395\n",
      "[262]\tvalidation_0-auc:0.93396\n",
      "[263]\tvalidation_0-auc:0.93395\n",
      "[264]\tvalidation_0-auc:0.93393\n",
      "[265]\tvalidation_0-auc:0.93391\n",
      "[266]\tvalidation_0-auc:0.93390\n",
      "[267]\tvalidation_0-auc:0.93394\n",
      "[268]\tvalidation_0-auc:0.93400\n",
      "[269]\tvalidation_0-auc:0.93399\n",
      "[270]\tvalidation_0-auc:0.93400\n",
      "[271]\tvalidation_0-auc:0.93399\n",
      "[272]\tvalidation_0-auc:0.93401\n",
      "[273]\tvalidation_0-auc:0.93409\n",
      "[274]\tvalidation_0-auc:0.93416\n",
      "[275]\tvalidation_0-auc:0.93417\n",
      "[276]\tvalidation_0-auc:0.93414\n",
      "[277]\tvalidation_0-auc:0.93412\n",
      "[278]\tvalidation_0-auc:0.93412\n",
      "[279]\tvalidation_0-auc:0.93408\n",
      "[280]\tvalidation_0-auc:0.93410\n",
      "[281]\tvalidation_0-auc:0.93410\n",
      "[282]\tvalidation_0-auc:0.93413\n",
      "[283]\tvalidation_0-auc:0.93410\n",
      "[284]\tvalidation_0-auc:0.93414\n",
      "[285]\tvalidation_0-auc:0.93415\n",
      "[286]\tvalidation_0-auc:0.93416\n",
      "[287]\tvalidation_0-auc:0.93416\n",
      "[288]\tvalidation_0-auc:0.93422\n",
      "[289]\tvalidation_0-auc:0.93420\n",
      "[290]\tvalidation_0-auc:0.93423\n",
      "[291]\tvalidation_0-auc:0.93423\n",
      "[292]\tvalidation_0-auc:0.93422\n",
      "[293]\tvalidation_0-auc:0.93417\n",
      "[294]\tvalidation_0-auc:0.93418\n",
      "[295]\tvalidation_0-auc:0.93421\n",
      "[296]\tvalidation_0-auc:0.93422\n",
      "[297]\tvalidation_0-auc:0.93428\n",
      "[298]\tvalidation_0-auc:0.93427\n",
      "[299]\tvalidation_0-auc:0.93430\n",
      "[300]\tvalidation_0-auc:0.93439\n",
      "[301]\tvalidation_0-auc:0.93438\n",
      "[302]\tvalidation_0-auc:0.93439\n",
      "[303]\tvalidation_0-auc:0.93439\n",
      "[304]\tvalidation_0-auc:0.93439\n",
      "[305]\tvalidation_0-auc:0.93439\n",
      "[306]\tvalidation_0-auc:0.93440\n",
      "[307]\tvalidation_0-auc:0.93443\n",
      "[308]\tvalidation_0-auc:0.93442\n",
      "[309]\tvalidation_0-auc:0.93441\n",
      "[310]\tvalidation_0-auc:0.93444\n",
      "[311]\tvalidation_0-auc:0.93444\n",
      "[312]\tvalidation_0-auc:0.93439\n",
      "[313]\tvalidation_0-auc:0.93437\n",
      "[314]\tvalidation_0-auc:0.93438\n",
      "[315]\tvalidation_0-auc:0.93437\n",
      "[316]\tvalidation_0-auc:0.93436\n",
      "[317]\tvalidation_0-auc:0.93438\n",
      "[318]\tvalidation_0-auc:0.93440\n",
      "[319]\tvalidation_0-auc:0.93441\n",
      "[320]\tvalidation_0-auc:0.93441\n",
      "[321]\tvalidation_0-auc:0.93440\n",
      "[322]\tvalidation_0-auc:0.93443\n",
      "[323]\tvalidation_0-auc:0.93449\n",
      "[324]\tvalidation_0-auc:0.93448\n",
      "[325]\tvalidation_0-auc:0.93448\n",
      "[326]\tvalidation_0-auc:0.93450\n",
      "[327]\tvalidation_0-auc:0.93450\n",
      "[328]\tvalidation_0-auc:0.93453\n",
      "[329]\tvalidation_0-auc:0.93447\n",
      "[330]\tvalidation_0-auc:0.93448\n",
      "[331]\tvalidation_0-auc:0.93449\n",
      "[332]\tvalidation_0-auc:0.93451\n",
      "[333]\tvalidation_0-auc:0.93451\n",
      "[334]\tvalidation_0-auc:0.93446\n",
      "[335]\tvalidation_0-auc:0.93446\n",
      "[336]\tvalidation_0-auc:0.93442\n",
      "[337]\tvalidation_0-auc:0.93443\n",
      "[338]\tvalidation_0-auc:0.93446\n",
      "[339]\tvalidation_0-auc:0.93443\n",
      "[340]\tvalidation_0-auc:0.93448\n",
      "[341]\tvalidation_0-auc:0.93449\n",
      "[342]\tvalidation_0-auc:0.93451\n",
      "[343]\tvalidation_0-auc:0.93450\n",
      "[344]\tvalidation_0-auc:0.93450\n",
      "[345]\tvalidation_0-auc:0.93460\n",
      "[346]\tvalidation_0-auc:0.93461\n",
      "[347]\tvalidation_0-auc:0.93462\n",
      "[348]\tvalidation_0-auc:0.93460\n",
      "[349]\tvalidation_0-auc:0.93462\n",
      "[350]\tvalidation_0-auc:0.93466\n",
      "[351]\tvalidation_0-auc:0.93465\n",
      "[352]\tvalidation_0-auc:0.93468\n",
      "[353]\tvalidation_0-auc:0.93468\n",
      "[354]\tvalidation_0-auc:0.93468\n",
      "[355]\tvalidation_0-auc:0.93468\n",
      "[356]\tvalidation_0-auc:0.93465\n",
      "[357]\tvalidation_0-auc:0.93466\n",
      "[358]\tvalidation_0-auc:0.93466\n",
      "[359]\tvalidation_0-auc:0.93465\n",
      "[360]\tvalidation_0-auc:0.93466\n",
      "[361]\tvalidation_0-auc:0.93466\n",
      "[362]\tvalidation_0-auc:0.93467\n",
      "[363]\tvalidation_0-auc:0.93466\n",
      "[364]\tvalidation_0-auc:0.93466\n",
      "[365]\tvalidation_0-auc:0.93466\n",
      "[366]\tvalidation_0-auc:0.93467\n",
      "[367]\tvalidation_0-auc:0.93466\n",
      "[368]\tvalidation_0-auc:0.93464\n",
      "[369]\tvalidation_0-auc:0.93464\n",
      "[370]\tvalidation_0-auc:0.93464\n",
      "[371]\tvalidation_0-auc:0.93464\n",
      "[372]\tvalidation_0-auc:0.93466\n",
      "[373]\tvalidation_0-auc:0.93464\n",
      "[374]\tvalidation_0-auc:0.93465\n",
      "[375]\tvalidation_0-auc:0.93466\n",
      "[376]\tvalidation_0-auc:0.93469\n",
      "[377]\tvalidation_0-auc:0.93467\n",
      "[378]\tvalidation_0-auc:0.93468\n",
      "[379]\tvalidation_0-auc:0.93467\n",
      "[380]\tvalidation_0-auc:0.93464\n",
      "[381]\tvalidation_0-auc:0.93466\n",
      "[382]\tvalidation_0-auc:0.93464\n",
      "[383]\tvalidation_0-auc:0.93464\n",
      "[384]\tvalidation_0-auc:0.93464\n",
      "[385]\tvalidation_0-auc:0.93462\n",
      "[386]\tvalidation_0-auc:0.93462\n",
      "[387]\tvalidation_0-auc:0.93463\n",
      "[388]\tvalidation_0-auc:0.93463\n",
      "[389]\tvalidation_0-auc:0.93463\n",
      "[390]\tvalidation_0-auc:0.93463\n",
      "[391]\tvalidation_0-auc:0.93463\n",
      "[392]\tvalidation_0-auc:0.93466\n",
      "[393]\tvalidation_0-auc:0.93465\n",
      "[394]\tvalidation_0-auc:0.93464\n",
      "[395]\tvalidation_0-auc:0.93463\n",
      "[396]\tvalidation_0-auc:0.93468\n",
      "[397]\tvalidation_0-auc:0.93470\n",
      "[398]\tvalidation_0-auc:0.93475\n",
      "[399]\tvalidation_0-auc:0.93476\n",
      "[400]\tvalidation_0-auc:0.93476\n",
      "[401]\tvalidation_0-auc:0.93475\n",
      "[402]\tvalidation_0-auc:0.93475\n",
      "[403]\tvalidation_0-auc:0.93472\n",
      "[404]\tvalidation_0-auc:0.93473\n",
      "[405]\tvalidation_0-auc:0.93474\n",
      "[406]\tvalidation_0-auc:0.93471\n",
      "[407]\tvalidation_0-auc:0.93470\n",
      "[408]\tvalidation_0-auc:0.93470\n",
      "[409]\tvalidation_0-auc:0.93474\n",
      "[410]\tvalidation_0-auc:0.93475\n",
      "[411]\tvalidation_0-auc:0.93477\n",
      "[412]\tvalidation_0-auc:0.93476\n",
      "[413]\tvalidation_0-auc:0.93476\n",
      "[414]\tvalidation_0-auc:0.93472\n",
      "[415]\tvalidation_0-auc:0.93471\n",
      "[416]\tvalidation_0-auc:0.93470\n",
      "[417]\tvalidation_0-auc:0.93470\n",
      "[418]\tvalidation_0-auc:0.93468\n",
      "[419]\tvalidation_0-auc:0.93468\n",
      "[420]\tvalidation_0-auc:0.93465\n",
      "[421]\tvalidation_0-auc:0.93465\n",
      "[422]\tvalidation_0-auc:0.93467\n",
      "[423]\tvalidation_0-auc:0.93467\n",
      "[424]\tvalidation_0-auc:0.93467\n",
      "[425]\tvalidation_0-auc:0.93468\n",
      "[426]\tvalidation_0-auc:0.93471\n",
      "[427]\tvalidation_0-auc:0.93471\n",
      "[428]\tvalidation_0-auc:0.93476\n",
      "[429]\tvalidation_0-auc:0.93477\n",
      "[430]\tvalidation_0-auc:0.93479\n",
      "[431]\tvalidation_0-auc:0.93481\n",
      "[432]\tvalidation_0-auc:0.93481\n",
      "[433]\tvalidation_0-auc:0.93480\n",
      "[434]\tvalidation_0-auc:0.93480\n",
      "[435]\tvalidation_0-auc:0.93481\n",
      "[436]\tvalidation_0-auc:0.93480\n",
      "[437]\tvalidation_0-auc:0.93484\n",
      "[438]\tvalidation_0-auc:0.93483\n",
      "[439]\tvalidation_0-auc:0.93480\n",
      "[440]\tvalidation_0-auc:0.93480\n",
      "[441]\tvalidation_0-auc:0.93476\n",
      "[442]\tvalidation_0-auc:0.93476\n",
      "[443]\tvalidation_0-auc:0.93474\n",
      "[444]\tvalidation_0-auc:0.93476\n",
      "[445]\tvalidation_0-auc:0.93475\n",
      "[446]\tvalidation_0-auc:0.93475\n",
      "[447]\tvalidation_0-auc:0.93477\n",
      "[448]\tvalidation_0-auc:0.93480\n",
      "[449]\tvalidation_0-auc:0.93484\n",
      "[450]\tvalidation_0-auc:0.93484\n",
      "[451]\tvalidation_0-auc:0.93486\n",
      "[452]\tvalidation_0-auc:0.93487\n",
      "[453]\tvalidation_0-auc:0.93486\n",
      "[454]\tvalidation_0-auc:0.93488\n",
      "[455]\tvalidation_0-auc:0.93491\n",
      "[456]\tvalidation_0-auc:0.93490\n",
      "[457]\tvalidation_0-auc:0.93489\n",
      "[458]\tvalidation_0-auc:0.93489\n",
      "[459]\tvalidation_0-auc:0.93487\n",
      "[460]\tvalidation_0-auc:0.93487\n",
      "[461]\tvalidation_0-auc:0.93487\n",
      "[462]\tvalidation_0-auc:0.93486\n",
      "[463]\tvalidation_0-auc:0.93485\n",
      "[464]\tvalidation_0-auc:0.93481\n",
      "[465]\tvalidation_0-auc:0.93482\n",
      "[466]\tvalidation_0-auc:0.93480\n",
      "[467]\tvalidation_0-auc:0.93480\n",
      "[468]\tvalidation_0-auc:0.93479\n",
      "[469]\tvalidation_0-auc:0.93478\n",
      "[470]\tvalidation_0-auc:0.93477\n",
      "[471]\tvalidation_0-auc:0.93474\n",
      "[472]\tvalidation_0-auc:0.93475\n",
      "[473]\tvalidation_0-auc:0.93473\n",
      "[474]\tvalidation_0-auc:0.93475\n",
      "[475]\tvalidation_0-auc:0.93475\n",
      "[476]\tvalidation_0-auc:0.93473\n",
      "[477]\tvalidation_0-auc:0.93471\n",
      "[478]\tvalidation_0-auc:0.93471\n",
      "[479]\tvalidation_0-auc:0.93470\n",
      "[480]\tvalidation_0-auc:0.93470\n",
      "[481]\tvalidation_0-auc:0.93470\n",
      "[482]\tvalidation_0-auc:0.93469\n",
      "[483]\tvalidation_0-auc:0.93470\n",
      "[484]\tvalidation_0-auc:0.93469\n",
      "[485]\tvalidation_0-auc:0.93467\n",
      "[486]\tvalidation_0-auc:0.93464\n",
      "[487]\tvalidation_0-auc:0.93465\n",
      "[488]\tvalidation_0-auc:0.93464\n",
      "[489]\tvalidation_0-auc:0.93464\n",
      "[490]\tvalidation_0-auc:0.93464\n",
      "[491]\tvalidation_0-auc:0.93464\n",
      "[492]\tvalidation_0-auc:0.93465\n",
      "[493]\tvalidation_0-auc:0.93466\n",
      "[494]\tvalidation_0-auc:0.93461\n",
      "[495]\tvalidation_0-auc:0.93462\n",
      "[496]\tvalidation_0-auc:0.93460\n",
      "[497]\tvalidation_0-auc:0.93467\n",
      "[498]\tvalidation_0-auc:0.93470\n",
      "[499]\tvalidation_0-auc:0.93469\n",
      "[500]\tvalidation_0-auc:0.93469\n",
      "[501]\tvalidation_0-auc:0.93469\n",
      "[502]\tvalidation_0-auc:0.93468\n",
      "[503]\tvalidation_0-auc:0.93468\n",
      "[504]\tvalidation_0-auc:0.93468\n",
      "[505]\tvalidation_0-auc:0.93471\n",
      "[506]\tvalidation_0-auc:0.93471\n",
      "[507]\tvalidation_0-auc:0.93470\n",
      "[508]\tvalidation_0-auc:0.93472\n",
      "[509]\tvalidation_0-auc:0.93471\n",
      "[510]\tvalidation_0-auc:0.93471\n",
      "[511]\tvalidation_0-auc:0.93471\n",
      "[512]\tvalidation_0-auc:0.93470\n",
      "[513]\tvalidation_0-auc:0.93470\n",
      "[514]\tvalidation_0-auc:0.93468\n",
      "[515]\tvalidation_0-auc:0.93469\n",
      "[516]\tvalidation_0-auc:0.93468\n",
      "[517]\tvalidation_0-auc:0.93465\n",
      "[518]\tvalidation_0-auc:0.93468\n",
      "[519]\tvalidation_0-auc:0.93467\n",
      "[520]\tvalidation_0-auc:0.93467\n",
      "[521]\tvalidation_0-auc:0.93466\n",
      "[522]\tvalidation_0-auc:0.93472\n",
      "[523]\tvalidation_0-auc:0.93470\n",
      "[524]\tvalidation_0-auc:0.93471\n",
      "[525]\tvalidation_0-auc:0.93471\n",
      "[526]\tvalidation_0-auc:0.93474\n",
      "[527]\tvalidation_0-auc:0.93474\n",
      "[528]\tvalidation_0-auc:0.93473\n",
      "[529]\tvalidation_0-auc:0.93476\n",
      "[530]\tvalidation_0-auc:0.93474\n",
      "[531]\tvalidation_0-auc:0.93474\n",
      "[532]\tvalidation_0-auc:0.93476\n",
      "[533]\tvalidation_0-auc:0.93476\n",
      "[534]\tvalidation_0-auc:0.93476\n",
      "[535]\tvalidation_0-auc:0.93476\n",
      "[536]\tvalidation_0-auc:0.93476\n",
      "[537]\tvalidation_0-auc:0.93476\n",
      "[538]\tvalidation_0-auc:0.93473\n",
      "[539]\tvalidation_0-auc:0.93474\n",
      "[540]\tvalidation_0-auc:0.93475\n",
      "[541]\tvalidation_0-auc:0.93473\n",
      "[542]\tvalidation_0-auc:0.93473\n",
      "[543]\tvalidation_0-auc:0.93473\n",
      "[544]\tvalidation_0-auc:0.93474\n",
      "[545]\tvalidation_0-auc:0.93475\n",
      "[546]\tvalidation_0-auc:0.93475\n",
      "[547]\tvalidation_0-auc:0.93474\n",
      "[548]\tvalidation_0-auc:0.93476\n",
      "[549]\tvalidation_0-auc:0.93475\n",
      "[550]\tvalidation_0-auc:0.93480\n",
      "[551]\tvalidation_0-auc:0.93480\n",
      "[552]\tvalidation_0-auc:0.93479\n",
      "[553]\tvalidation_0-auc:0.93478\n",
      "[554]\tvalidation_0-auc:0.93480\n",
      "[555]\tvalidation_0-auc:0.93480\n",
      "Stopping. Best iteration:\n",
      "[455]\tvalidation_0-auc:0.93491\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Hyperparameter tuning and cross validation\n",
    "\n",
    "#tune_model2 = grid_search2.fit(X = x2_train,\n",
    "                             #y = y2_train,\n",
    "                             #**evaluation_parameters)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "thousand-granny",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'booster': 'gbtree', 'colsample_bytree': 0.5, 'gamma': 0.1, 'learning_rate': 0.05, 'max_depth': 6, 'min_child_weight': 2, 'n_estimators': 600, 'random_state': 1502, 'subsample': 1} 0.9355959064807038\n"
     ]
    }
   ],
   "source": [
    "print(grid_search2.best_params_, grid_search2.best_score_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "destroyed-thailand",
   "metadata": {},
   "source": [
    "### 3rd hyper parameter tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "pending-coral",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Set the parameters part 3\n",
    "\n",
    "parameters3 = {'learning_rate': 0.05,\n",
    "               'max_depth': 6,\n",
    "               'colsample_bytree': 0.5,\n",
    "               'subsample': 1,\n",
    "               'min_child_weight': 2,\n",
    "               'gamma': 0, \n",
    "               'random_state': 1502,\n",
    "               'eval_metric': \"auc\",\n",
    "               'objective': \"binary:logistic\"}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "preliminary-scholar",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0]\tYes-auc:0.85589\n",
      "[50]\tYes-auc:0.92302\n",
      "[100]\tYes-auc:0.92802\n",
      "[150]\tYes-auc:0.93209\n",
      "[200]\tYes-auc:0.93345\n",
      "[250]\tYes-auc:0.93371\n",
      "[300]\tYes-auc:0.93415\n",
      "[350]\tYes-auc:0.93449\n",
      "[400]\tYes-auc:0.93475\n",
      "[450]\tYes-auc:0.93513\n",
      "[500]\tYes-auc:0.93504\n",
      "[550]\tYes-auc:0.93505\n",
      "[600]\tYes-auc:0.93515\n",
      "[650]\tYes-auc:0.93495\n",
      "[700]\tYes-auc:0.93479\n",
      "[750]\tYes-auc:0.93479\n",
      "[799]\tYes-auc:0.93477\n"
     ]
    }
   ],
   "source": [
    "# Run XGBoost part 3\n",
    "model3 = xgb.train(params = parameters3,\n",
    "                   dtrain = Train_2,\n",
    "                   num_boost_round = 800,\n",
    "                   evals = [(Test_2, \"Yes\")],\n",
    "                   verbose_eval = 50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "discrete-conditioning",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Predictions part 3\n",
    "predictions3 = model3.predict(Test_2)\n",
    "predictions3 = np.where(predictions3 > 0.05, 1, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "concerned-movement",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "accuracy of the train data set 0.9085480482140883\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.99      0.76      0.86      7993\n",
      "           1       0.34      0.96      0.51      1050\n",
      "\n",
      "    accuracy                           0.78      9043\n",
      "   macro avg       0.67      0.86      0.68      9043\n",
      "weighted avg       0.92      0.78      0.82      9043\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Confusion Matrix\n",
    "confusion_matrix3 = confusion_matrix(y2_test, predictions3)\n",
    "print(\"accuracy of the train data set\" + \" \"+ str(accuracy_score(y2_test,predictions2)))\n",
    "report3 = classification_report(y_test, predictions3)\n",
    "print(report3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "assigned-gasoline",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(69.0, 0.5, 'Truth')"
      ]
     },
     "execution_count": 86,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10,8))\n",
    "sns.heatmap(confusion_matrix3,annot=True,fmt='d')\n",
    "plt.xlabel('Predicted')\n",
    "plt.ylabel('Truth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "bearing-rotation",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'Feature importance'}, xlabel='F score', ylabel='Features'>"
      ]
     },
     "execution_count": 57,
     "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": [
    "#plot importances\n",
    "xgb.plot_importance(model3, max_num_features = 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "departmental-amount",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Preparing SHAP\n",
    "#pip install shap\n",
    "#conda install -c conda-forge shap\n",
    "import shap\n",
    "explainer = shap.TreeExplainer(model3)\n",
    "shap_values = explainer.shap_values(x2_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "threatened-height",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x396 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#shap plotting\n",
    "shap.summary_plot(shap_values,\n",
    "                  x2_test,\n",
    "                  feature_names = feature_columns,\n",
    "                  max_display = 10)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "numeric-surfing",
   "metadata": {},
   "source": [
    "### Build ANN with one hidden layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "english-solid",
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import Sequential\n",
    "from tensorflow.keras.layers import Flatten, Dense\n",
    "import keras\n",
    "from keras import layers\n",
    "from keras import models\n",
    "from keras import utils\n",
    "from keras.layers import Dense\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Flatten\n",
    "from keras.layers import Dropout\n",
    "from keras.layers import Activation\n",
    "from keras.regularizers import l2\n",
    "from keras.optimizers import SGD\n",
    "from keras.optimizers import RMSprop\n",
    "from keras import datasets\n",
    "\n",
    "from keras.callbacks import LearningRateScheduler\n",
    "from keras.callbacks import History\n",
    "\n",
    "from keras import losses\n",
    "from sklearn.utils import shuffle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "artistic-briefing",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_ann=final_df.iloc[:,0:42]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "reverse-textbook",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>age</th>\n",
       "      <th>balance</th>\n",
       "      <th>day</th>\n",
       "      <th>duration</th>\n",
       "      <th>campaign</th>\n",
       "      <th>pdays</th>\n",
       "      <th>previous</th>\n",
       "      <th>job_blue-collar</th>\n",
       "      <th>job_entrepreneur</th>\n",
       "      <th>job_housemaid</th>\n",
       "      <th>...</th>\n",
       "      <th>month_jun</th>\n",
       "      <th>month_mar</th>\n",
       "      <th>month_may</th>\n",
       "      <th>month_nov</th>\n",
       "      <th>month_oct</th>\n",
       "      <th>month_sep</th>\n",
       "      <th>poutcome_other</th>\n",
       "      <th>poutcome_success</th>\n",
       "      <th>poutcome_unknown</th>\n",
       "      <th>y_yes</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>58</td>\n",
       "      <td>2143</td>\n",
       "      <td>5</td>\n",
       "      <td>261</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>44</td>\n",
       "      <td>29</td>\n",
       "      <td>5</td>\n",
       "      <td>151</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>33</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>76</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>47</td>\n",
       "      <td>1506</td>\n",
       "      <td>5</td>\n",
       "      <td>92</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>33</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>198</td>\n",
       "      <td>1</td>\n",
       "      <td>-1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 43 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   age  balance  day  duration  campaign  pdays  previous  job_blue-collar  \\\n",
       "0   58     2143    5       261         1     -1         0                0   \n",
       "1   44       29    5       151         1     -1         0                0   \n",
       "2   33        2    5        76         1     -1         0                0   \n",
       "3   47     1506    5        92         1     -1         0                1   \n",
       "4   33        1    5       198         1     -1         0                0   \n",
       "\n",
       "   job_entrepreneur  job_housemaid  ...  month_jun  month_mar  month_may  \\\n",
       "0                 0              0  ...          0          0          1   \n",
       "1                 0              0  ...          0          0          1   \n",
       "2                 1              0  ...          0          0          1   \n",
       "3                 0              0  ...          0          0          1   \n",
       "4                 0              0  ...          0          0          1   \n",
       "\n",
       "   month_nov  month_oct  month_sep  poutcome_other  poutcome_success  \\\n",
       "0          0          0          0               0                 0   \n",
       "1          0          0          0               0                 0   \n",
       "2          0          0          0               0                 0   \n",
       "3          0          0          0               0                 0   \n",
       "4          0          0          0               0                 0   \n",
       "\n",
       "   poutcome_unknown  y_yes  \n",
       "0                 1      0  \n",
       "1                 1      0  \n",
       "2                 1      0  \n",
       "3                 1      0  \n",
       "4                 1      0  \n",
       "\n",
       "[5 rows x 43 columns]"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "final_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "headed-preview",
   "metadata": {},
   "source": [
    "### Feature Scaling Process"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "intense-pastor",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.model_selection import train_test_split\n",
    "sc=StandardScaler()\n",
    "x_train,x_test,y_train,y_test= train_test_split(x_ann, y, test_size = 0.2, random_state = 0, stratify = y)\n",
    "x_train=sc.fit_transform(x_train)\n",
    "x_test=sc.fit_transform(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "hundred-luther",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(36168, 42)\n"
     ]
    }
   ],
   "source": [
    "print(x_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "supposed-match",
   "metadata": {},
   "outputs": [],
   "source": [
    "model=Sequential()\n",
    "# Firstly, we will develop the input layer \n",
    "model.add(Dense(x_ann.shape[1],activation=\"relu\",kernel_initializer = 'he_normal',bias_initializer='zeros'))\n",
    "# Hidden Layer\n",
    "model.add(Dense(128,activation=\"relu\"))\n",
    "#Out put layer\n",
    "model.add(Dense(1,activation=\"sigmoid\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "tutorial-disposal",
   "metadata": {},
   "source": [
    "### Compiling the ANN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "suspended-township",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(optimizer=\"adamax\",loss='binary_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "vital-integer",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.2651 - accuracy: 0.8941\n",
      "Epoch 2/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.2236 - accuracy: 0.9018\n",
      "Epoch 3/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.2146 - accuracy: 0.9038\n",
      "Epoch 4/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.2102 - accuracy: 0.9062\n",
      "Epoch 5/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.2064 - accuracy: 0.9069\n",
      "Epoch 6/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.2031 - accuracy: 0.9094\n",
      "Epoch 7/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.2008 - accuracy: 0.9098\n",
      "Epoch 8/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.1987 - accuracy: 0.9101\n",
      "Epoch 9/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1965 - accuracy: 0.9112\n",
      "Epoch 10/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1947 - accuracy: 0.9125\n",
      "Epoch 11/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.1929 - accuracy: 0.9137\n",
      "Epoch 12/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.1912 - accuracy: 0.9135\n",
      "Epoch 13/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.1899 - accuracy: 0.9151\n",
      "Epoch 14/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1882 - accuracy: 0.9157\n",
      "Epoch 15/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1871 - accuracy: 0.9167\n",
      "Epoch 16/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.1858 - accuracy: 0.9172\n",
      "Epoch 17/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1845 - accuracy: 0.9184\n",
      "Epoch 18/20\n",
      "3617/3617 [==============================] - 7s 2ms/step - loss: 0.1830 - accuracy: 0.9189\n",
      "Epoch 19/20\n",
      "3617/3617 [==============================] - 6s 2ms/step - loss: 0.1820 - accuracy: 0.9195\n",
      "Epoch 20/20\n",
      "3617/3617 [==============================] - 4s 1ms/step - loss: 0.1812 - accuracy: 0.9182\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x2923615fbe0>"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(x=x_train,y=y_train,batch_size=10,epochs=20,verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "unknown-survivor",
   "metadata": {},
   "source": [
    "### Build ANN with two hidden layers and different optimizers function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "id": "beginning-treasurer",
   "metadata": {},
   "outputs": [],
   "source": [
    "initializer = tf.keras.initializers.VarianceScaling(scale=0.1, mode='fan_in', distribution='uniform')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "sexual-buying",
   "metadata": {},
   "outputs": [],
   "source": [
    "model2=Sequential()\n",
    "# Firstly, we will develop the input layer \n",
    "model2.add(Dense(x_ann.shape[1],activation=\"relu\",\n",
    "                 kernel_initializer = initializer,bias_initializer='zeros',input_dim = x_ann.shape[1]))\n",
    "# first Hidden Layer\n",
    "model2.add(Dense(128,activation=\"relu\",kernel_initializer = initializer))\n",
    "#Second hidden layer\n",
    "model2.add(Dense(64,activation=\"relu\",kernel_initializer = initializer))\n",
    "#Out put layer\n",
    "model2.add(Dense(1,activation=\"sigmoid\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "muslim-tissue",
   "metadata": {},
   "outputs": [],
   "source": [
    "model2.compile(optimizer=\"adamax\",loss='binary_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "herbal-riding",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1724 - accuracy: 0.9229\n",
      "Epoch 2/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1708 - accuracy: 0.9247\n",
      "Epoch 3/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1697 - accuracy: 0.9249\n",
      "Epoch 4/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1683 - accuracy: 0.9253\n",
      "Epoch 5/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1670 - accuracy: 0.9276\n",
      "Epoch 6/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1653 - accuracy: 0.9285\n",
      "Epoch 7/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1638 - accuracy: 0.9278\n",
      "Epoch 8/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1625 - accuracy: 0.9309\n",
      "Epoch 9/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1618 - accuracy: 0.9309\n",
      "Epoch 10/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1599 - accuracy: 0.9312\n",
      "Epoch 11/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1580 - accuracy: 0.9321\n",
      "Epoch 12/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1567 - accuracy: 0.9318\n",
      "Epoch 13/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1558 - accuracy: 0.9341\n",
      "Epoch 14/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1543 - accuracy: 0.9347\n",
      "Epoch 15/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1531 - accuracy: 0.9351\n",
      "Epoch 16/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1517 - accuracy: 0.9358\n",
      "Epoch 17/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1501 - accuracy: 0.9348\n",
      "Epoch 18/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1490 - accuracy: 0.9373\n",
      "Epoch 19/20\n",
      "3617/3617 [==============================] - 5s 1ms/step - loss: 0.1472 - accuracy: 0.9379\n",
      "Epoch 20/20\n",
      "3617/3617 [==============================] - 5s 2ms/step - loss: 0.1462 - accuracy: 0.9388\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x2923b5110a0>"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model2.fit(x=x_train,y=y_train,batch_size=10,epochs=20,verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "numeric-attraction",
   "metadata": {},
   "source": [
    "### Tune the hyper parameters of DNN parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "major-external",
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.wrappers.scikit_learn import KerasClassifier\n",
    "from sklearn.model_selection import cross_val_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "id": "absolute-enhancement",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Assign some values for learningrate, epochs and decay_rate\n",
    "epochs=60\n",
    "learning_rate = 0.1\n",
    "decay_rate = learning_rate / epochs\n",
    "momentum = 0.8\n",
    "# Stochastic gradeint desent\n",
    "sgd = SGD(lr=learning_rate, momentum=momentum, decay=decay_rate, nesterov=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "id": "collected-mounting",
   "metadata": {},
   "outputs": [],
   "source": [
    "# build the model\n",
    "input_dim = x_ann.shape[1]\n",
    "lr_model = Sequential()\n",
    "lr_model.add(Dense(x_ann.shape[1], activation=tf.nn.relu, kernel_initializer='uniform', \n",
    "                input_dim = input_dim)) \n",
    "lr_model.add(Dropout(0.1))\n",
    "lr_model.add(Dense(128, kernel_initializer='uniform', activation=tf.nn.relu))\n",
    "lr_model.add(Dense(1, kernel_initializer='uniform', activation=tf.nn.sigmoid))\n",
    "\n",
    "# compile the model\n",
    "lr_model.compile(loss='binary_crossentropy',\n",
    "              optimizer=sgd,\n",
    "              metrics=['acc'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "id": "important-florist",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.2382 - acc: 0.8969 - val_loss: 0.2140 - val_acc: 0.9025\n",
      "Epoch 2/60\n",
      "1131/1131 [==============================] - 1s 882us/step - loss: 0.2122 - acc: 0.9035 - val_loss: 0.2111 - val_acc: 0.9037\n",
      "Epoch 3/60\n",
      "1131/1131 [==============================] - 1s 909us/step - loss: 0.2075 - acc: 0.9042 - val_loss: 0.2076 - val_acc: 0.9039\n",
      "Epoch 4/60\n",
      "1131/1131 [==============================] - 1s 885us/step - loss: 0.2040 - acc: 0.9056 - val_loss: 0.2071 - val_acc: 0.9040\n",
      "Epoch 5/60\n",
      "1131/1131 [==============================] - 1s 921us/step - loss: 0.2024 - acc: 0.9067 - val_loss: 0.2063 - val_acc: 0.9028\n",
      "Epoch 6/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.2003 - acc: 0.9075 - val_loss: 0.2042 - val_acc: 0.9042\n",
      "Epoch 7/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1984 - acc: 0.9083 - val_loss: 0.2051 - val_acc: 0.9049\n",
      "Epoch 8/60\n",
      "1131/1131 [==============================] - 1s 919us/step - loss: 0.1980 - acc: 0.9091 - val_loss: 0.2044 - val_acc: 0.9043\n",
      "Epoch 9/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1960 - acc: 0.9094 - val_loss: 0.2041 - val_acc: 0.9052\n",
      "Epoch 10/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1954 - acc: 0.9102 - val_loss: 0.2029 - val_acc: 0.9066\n",
      "Epoch 11/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1948 - acc: 0.9109 - val_loss: 0.2037 - val_acc: 0.9046\n",
      "Epoch 12/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1938 - acc: 0.9106 - val_loss: 0.2030 - val_acc: 0.9041\n",
      "Epoch 13/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1936 - acc: 0.9124 - val_loss: 0.2033 - val_acc: 0.9036\n",
      "Epoch 14/60\n",
      "1131/1131 [==============================] - 2s 2ms/step - loss: 0.1930 - acc: 0.9112 - val_loss: 0.2031 - val_acc: 0.9046\n",
      "Epoch 15/60\n",
      "1131/1131 [==============================] - 1s 954us/step - loss: 0.1921 - acc: 0.9120 - val_loss: 0.2023 - val_acc: 0.9056\n",
      "Epoch 16/60\n",
      "1131/1131 [==============================] - 1s 899us/step - loss: 0.1920 - acc: 0.9123 - val_loss: 0.2023 - val_acc: 0.9061\n",
      "Epoch 17/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1916 - acc: 0.9124 - val_loss: 0.2022 - val_acc: 0.9057\n",
      "Epoch 18/60\n",
      "1131/1131 [==============================] - 1s 887us/step - loss: 0.1913 - acc: 0.9122 - val_loss: 0.2025 - val_acc: 0.9056\n",
      "Epoch 19/60\n",
      "1131/1131 [==============================] - 1s 934us/step - loss: 0.1907 - acc: 0.9121 - val_loss: 0.2020 - val_acc: 0.9066\n",
      "Epoch 20/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1902 - acc: 0.9127 - val_loss: 0.2029 - val_acc: 0.9055\n",
      "Epoch 21/60\n",
      "1131/1131 [==============================] - 1s 959us/step - loss: 0.1897 - acc: 0.9132 - val_loss: 0.2032 - val_acc: 0.9053\n",
      "Epoch 22/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1897 - acc: 0.9131 - val_loss: 0.2026 - val_acc: 0.9064\n",
      "Epoch 23/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1900 - acc: 0.9130 - val_loss: 0.2027 - val_acc: 0.9062\n",
      "Epoch 24/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1881 - acc: 0.9139 - val_loss: 0.2030 - val_acc: 0.9063\n",
      "Epoch 25/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1895 - acc: 0.9134 - val_loss: 0.2030 - val_acc: 0.9062\n",
      "Epoch 26/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1897 - acc: 0.9133 - val_loss: 0.2028 - val_acc: 0.9078\n",
      "Epoch 27/60\n",
      "1131/1131 [==============================] - 2s 1ms/step - loss: 0.1884 - acc: 0.9136 - val_loss: 0.2029 - val_acc: 0.9071\n",
      "Epoch 28/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1876 - acc: 0.9152 - val_loss: 0.2028 - val_acc: 0.9059\n",
      "Epoch 29/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1879 - acc: 0.9149 - val_loss: 0.2025 - val_acc: 0.9067\n",
      "Epoch 30/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1875 - acc: 0.9151 - val_loss: 0.2026 - val_acc: 0.9062\n",
      "Epoch 31/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1882 - acc: 0.9135 - val_loss: 0.2029 - val_acc: 0.9058\n",
      "Epoch 32/60\n",
      "1131/1131 [==============================] - 2s 2ms/step - loss: 0.1884 - acc: 0.9136 - val_loss: 0.2025 - val_acc: 0.9057\n",
      "Epoch 33/60\n",
      "1131/1131 [==============================] - 2s 1ms/step - loss: 0.1865 - acc: 0.9150 - val_loss: 0.2031 - val_acc: 0.9056\n",
      "Epoch 34/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1868 - acc: 0.9143 - val_loss: 0.2037 - val_acc: 0.9060\n",
      "Epoch 35/60\n",
      "1131/1131 [==============================] - 2s 1ms/step - loss: 0.1871 - acc: 0.9155 - val_loss: 0.2031 - val_acc: 0.9070\n",
      "Epoch 36/60\n",
      "1131/1131 [==============================] - 2s 1ms/step - loss: 0.1870 - acc: 0.9152 - val_loss: 0.2029 - val_acc: 0.9066\n",
      "Epoch 37/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1870 - acc: 0.9138 - val_loss: 0.2028 - val_acc: 0.9072\n",
      "Epoch 38/60\n",
      "1131/1131 [==============================] - 2s 1ms/step - loss: 0.1860 - acc: 0.9151 - val_loss: 0.2029 - val_acc: 0.9072\n",
      "Epoch 39/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1861 - acc: 0.9152 - val_loss: 0.2030 - val_acc: 0.9066\n",
      "Epoch 40/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1858 - acc: 0.9144 - val_loss: 0.2035 - val_acc: 0.9061\n",
      "Epoch 41/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1857 - acc: 0.9148 - val_loss: 0.2032 - val_acc: 0.9069\n",
      "Epoch 42/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1865 - acc: 0.9151 - val_loss: 0.2035 - val_acc: 0.9064\n",
      "Epoch 43/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1871 - acc: 0.9146 - val_loss: 0.2033 - val_acc: 0.9062\n",
      "Epoch 44/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1852 - acc: 0.9160 - val_loss: 0.2032 - val_acc: 0.9061\n",
      "Epoch 45/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1857 - acc: 0.9160 - val_loss: 0.2032 - val_acc: 0.9051\n",
      "Epoch 46/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1862 - acc: 0.9143 - val_loss: 0.2031 - val_acc: 0.9070\n",
      "Epoch 47/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1847 - acc: 0.9159 - val_loss: 0.2035 - val_acc: 0.9060\n",
      "Epoch 48/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1854 - acc: 0.9147 - val_loss: 0.2036 - val_acc: 0.9059\n",
      "Epoch 49/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1854 - acc: 0.9157 - val_loss: 0.2036 - val_acc: 0.9062\n",
      "Epoch 50/60\n",
      "1131/1131 [==============================] - 2s 1ms/step - loss: 0.1838 - acc: 0.9165 - val_loss: 0.2037 - val_acc: 0.9061\n",
      "Epoch 51/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1851 - acc: 0.9152 - val_loss: 0.2035 - val_acc: 0.9070\n",
      "Epoch 52/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1855 - acc: 0.9161 - val_loss: 0.2037 - val_acc: 0.9061\n",
      "Epoch 53/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1853 - acc: 0.9157 - val_loss: 0.2034 - val_acc: 0.9071\n",
      "Epoch 54/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1850 - acc: 0.9164 - val_loss: 0.2034 - val_acc: 0.9068\n",
      "Epoch 55/60\n",
      "1131/1131 [==============================] - 1s 991us/step - loss: 0.1845 - acc: 0.9154 - val_loss: 0.2034 - val_acc: 0.9064\n",
      "Epoch 56/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1848 - acc: 0.9155 - val_loss: 0.2037 - val_acc: 0.9067\n",
      "Epoch 57/60\n",
      "1131/1131 [==============================] - 1s 1000us/step - loss: 0.1845 - acc: 0.9158 - val_loss: 0.2040 - val_acc: 0.9060\n",
      "Epoch 58/60\n",
      "1131/1131 [==============================] - 1s 984us/step - loss: 0.1847 - acc: 0.9155 - val_loss: 0.2036 - val_acc: 0.9069\n",
      "Epoch 59/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1842 - acc: 0.9158 - val_loss: 0.2034 - val_acc: 0.9070\n",
      "Epoch 60/60\n",
      "1131/1131 [==============================] - 1s 1ms/step - loss: 0.1839 - acc: 0.9154 - val_loss: 0.2039 - val_acc: 0.9064\n",
      "Wall time: 1min 17s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# Fit the model\n",
    "batch_size = int(input_dim/100)\n",
    "\n",
    "lr_model_history = lr_model.fit(x_train, y_train,\n",
    "                    batch_size=batch_size,\n",
    "                    epochs=epochs,\n",
    "                    verbose=1,\n",
    "                    validation_data=(x_test, y_test))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "loaded-assessment",
   "metadata": {},
   "source": [
    "### Plot the loss vs epoch graph"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "id": "integrated-subcommittee",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the loss function\n",
    "fig, ax = plt.subplots(1, 1, figsize=(10,6))\n",
    "ax.plot(np.sqrt(lr_model_history.history['loss']), 'r', label='train')\n",
    "ax.plot(np.sqrt(lr_model_history.history['val_loss']), 'b' ,label='val')\n",
    "ax.set_xlabel(r'Epoch', fontsize=20)\n",
    "ax.set_ylabel(r'Loss', fontsize=20)\n",
    "ax.legend()\n",
    "ax.tick_params(labelsize=20)\n",
    "\n",
    "# Plot the accuracy\n",
    "fig, ax = plt.subplots(1, 1, figsize=(10,6))\n",
    "ax.plot(np.sqrt(lr_model_history.history['acc']), 'r', label='train')\n",
    "ax.plot(np.sqrt(lr_model_history.history['val_acc']), 'b' ,label='val')\n",
    "ax.set_xlabel(r'Epoch', fontsize=20)\n",
    "ax.set_ylabel(r'Accuracy', fontsize=20)\n",
    "ax.legend()\n",
    "ax.tick_params(labelsize=20)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "contemporary-coalition",
   "metadata": {},
   "source": [
    "### Develop the exponential decay model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "id": "local-telescope",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3756 - acc: 0.8829 - val_loss: 0.3569 - val_acc: 0.8830\n",
      "Epoch 2/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3553 - acc: 0.8830 - val_loss: 0.3539 - val_acc: 0.8830\n",
      "Epoch 3/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3530 - acc: 0.8830 - val_loss: 0.3520 - val_acc: 0.8830\n",
      "Epoch 4/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3514 - acc: 0.8830 - val_loss: 0.3507 - val_acc: 0.8830\n",
      "Epoch 5/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3502 - acc: 0.8830 - val_loss: 0.3496 - val_acc: 0.8830\n",
      "Epoch 6/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3493 - acc: 0.8830 - val_loss: 0.3487 - val_acc: 0.8830\n",
      "Epoch 7/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3484 - acc: 0.8830 - val_loss: 0.3480 - val_acc: 0.8830\n",
      "Epoch 8/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3478 - acc: 0.8830 - val_loss: 0.3474 - val_acc: 0.8830\n",
      "Epoch 9/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3473 - acc: 0.8830 - val_loss: 0.3469 - val_acc: 0.8830\n",
      "Epoch 10/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3469 - acc: 0.8830 - val_loss: 0.3465 - val_acc: 0.8830\n",
      "Epoch 11/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3464 - acc: 0.8830 - val_loss: 0.3462 - val_acc: 0.8830\n",
      "Epoch 12/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3462 - acc: 0.8830 - val_loss: 0.3459 - val_acc: 0.8830\n",
      "Epoch 13/60\n",
      "185/185 [==============================] - 0s 3ms/step - loss: 0.3459 - acc: 0.8830 - val_loss: 0.3456 - val_acc: 0.8830\n",
      "Epoch 14/60\n",
      "185/185 [==============================] - 0s 3ms/step - loss: 0.3457 - acc: 0.8830 - val_loss: 0.3454 - val_acc: 0.8830\n",
      "Epoch 15/60\n",
      "185/185 [==============================] - 0s 3ms/step - loss: 0.3455 - acc: 0.8830 - val_loss: 0.3452 - val_acc: 0.8830\n",
      "Epoch 16/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3453 - acc: 0.8830 - val_loss: 0.3451 - val_acc: 0.8830\n",
      "Epoch 17/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3451 - acc: 0.8830 - val_loss: 0.3449 - val_acc: 0.8830\n",
      "Epoch 18/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3450 - acc: 0.8830 - val_loss: 0.3448 - val_acc: 0.8830\n",
      "Epoch 19/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3449 - acc: 0.8830 - val_loss: 0.3447 - val_acc: 0.8830\n",
      "Epoch 20/60\n",
      "185/185 [==============================] - 0s 3ms/step - loss: 0.3449 - acc: 0.8830 - val_loss: 0.3447 - val_acc: 0.8830\n",
      "Epoch 21/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3448 - acc: 0.8830 - val_loss: 0.3446 - val_acc: 0.8830\n",
      "Epoch 22/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3447 - acc: 0.8830 - val_loss: 0.3445 - val_acc: 0.8830\n",
      "Epoch 23/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3446 - acc: 0.8830 - val_loss: 0.3445 - val_acc: 0.8830\n",
      "Epoch 24/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3446 - acc: 0.8830 - val_loss: 0.3444 - val_acc: 0.8830\n",
      "Epoch 25/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3445 - acc: 0.8830 - val_loss: 0.3444 - val_acc: 0.8830\n",
      "Epoch 26/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3445 - acc: 0.8830 - val_loss: 0.3443 - val_acc: 0.8830\n",
      "Epoch 27/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3445 - acc: 0.8830 - val_loss: 0.3443 - val_acc: 0.8830\n",
      "Epoch 28/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3445 - acc: 0.8830 - val_loss: 0.3443 - val_acc: 0.8830\n",
      "Epoch 29/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3444 - acc: 0.8830 - val_loss: 0.3443 - val_acc: 0.8830\n",
      "Epoch 30/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3444 - acc: 0.8830 - val_loss: 0.3442 - val_acc: 0.8830\n",
      "Epoch 31/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3442 - val_acc: 0.8830\n",
      "Epoch 32/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3444 - acc: 0.8830 - val_loss: 0.3442 - val_acc: 0.8830\n",
      "Epoch 33/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3444 - acc: 0.8830 - val_loss: 0.3442 - val_acc: 0.8830\n",
      "Epoch 34/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3442 - val_acc: 0.8830\n",
      "Epoch 35/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3442 - val_acc: 0.8830\n",
      "Epoch 36/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3444 - acc: 0.8830 - val_loss: 0.3442 - val_acc: 0.8830\n",
      "Epoch 37/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 38/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 39/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 40/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 41/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3442 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 42/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 43/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 44/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 45/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 46/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 47/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 48/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 49/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 50/60\n",
      "185/185 [==============================] - 0s 3ms/step - loss: 0.3442 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 51/60\n",
      "185/185 [==============================] - 0s 3ms/step - loss: 0.3442 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 52/60\n",
      "185/185 [==============================] - 0s 3ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 53/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 54/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3442 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 55/60\n",
      "185/185 [==============================] - 0s 3ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 56/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3442 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 57/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 58/60\n",
      "185/185 [==============================] - 0s 3ms/step - loss: 0.3443 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 59/60\n",
      "185/185 [==============================] - 0s 3ms/step - loss: 0.3442 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n",
      "Epoch 60/60\n",
      "185/185 [==============================] - 0s 2ms/step - loss: 0.3442 - acc: 0.8830 - val_loss: 0.3441 - val_acc: 0.8830\n"
     ]
    }
   ],
   "source": [
    "\n",
    "epochs = 60\n",
    "learning_rate = 0.1 # initial learning rate\n",
    "decay_rate = 0.1\n",
    "momentum = 0.8\n",
    "\n",
    "# define the optimizer function\n",
    "sgd = SGD(lr=learning_rate, momentum=momentum, decay=decay_rate, nesterov=False)\n",
    "\n",
    "input_dim = x_ann.shape[1]\n",
    "num_classes = 1\n",
    "batch_size = 196\n",
    "\n",
    "# build the model\n",
    "exponential_decay_model = Sequential()\n",
    "exponential_decay_model.add(Dense(x_ann.shape[1], activation=tf.nn.relu, kernel_initializer='uniform', \n",
    "                                  input_dim = input_dim))\n",
    "exponential_decay_model.add(Dropout(0.1))\n",
    "exponential_decay_model.add(Dense(128, kernel_initializer='uniform', activation=tf.nn.relu))\n",
    "exponential_decay_model.add(Dense(num_classes, kernel_initializer='uniform', activation=tf.nn.sigmoid))\n",
    "\n",
    "# compile the model\n",
    "exponential_decay_model.compile(loss='binary_crossentropy', \n",
    "                                optimizer=sgd, \n",
    "                                metrics=['acc'])\n",
    "                                \n",
    "# define the learning rate change \n",
    "def exp_decay(epoch):\n",
    "    lrate = learning_rate * np.exp(-decay_rate*epoch)\n",
    "    return lrate\n",
    "    \n",
    "# learning schedule callback\n",
    "loss_history = History()\n",
    "lr_rate = LearningRateScheduler(exp_decay)\n",
    "callbacks_list = [loss_history, lr_rate]\n",
    "\n",
    "# you invoke the LearningRateScheduler during the .fit() phase\n",
    "exponential_decay_model_history = exponential_decay_model.fit(x_train, y_train,\n",
    "                                    batch_size=batch_size,\n",
    "                                    epochs=epochs,\n",
    "                                    callbacks=callbacks_list,\n",
    "                                    verbose=1,\n",
    "                                    validation_data=(x_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "id": "identified-cigarette",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Develop the model and compile\n",
    "\n",
    "model = Sequential()\n",
    "model.add(Dense(input_dim, activation=tf.nn.relu, kernel_initializer='uniform', \n",
    "                input_dim = input_dim)) # fully-connected layer with 64 hidden units\n",
    "model.add(Dropout(0.1))\n",
    "model.add(Dense(128, kernel_initializer='uniform', activation=tf.nn.relu))\n",
    "model.add(Dense(num_classes, kernel_initializer='uniform', activation=tf.nn.sigmoid))\n",
    "\n",
    "# defining the parameters for RMSprop (I used the keras defaults here)\n",
    "rms = RMSprop(lr=0.001, rho=0.9, epsilon=None, decay=0.0)\n",
    "\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer=rms,\n",
    "              metrics=['acc'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "id": "mobile-persian",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7189 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 2/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7181 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 3/60\n",
      "862/862 [==============================] - 1s 977us/step - loss: 1.3945e-08 - acc: 0.7186 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 4/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7167 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 5/60\n",
      "862/862 [==============================] - 1s 984us/step - loss: 1.3945e-08 - acc: 0.7200 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 6/60\n",
      "862/862 [==============================] - 1s 999us/step - loss: 1.3945e-08 - acc: 0.7193 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 7/60\n",
      "862/862 [==============================] - 1s 982us/step - loss: 1.3945e-08 - acc: 0.7213 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 8/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7190 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 9/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7178 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 10/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7175 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 11/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7194 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 12/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7189 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 13/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7183 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 14/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7192 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 15/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7207 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 16/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7193 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 17/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7186 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 18/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7184 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 19/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7178 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 20/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7175 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 21/60\n",
      "862/862 [==============================] - 1s 2ms/step - loss: 1.3945e-08 - acc: 0.7202 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 22/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7184 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 23/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7193 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 24/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7202 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 25/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7185 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 26/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7200 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 27/60\n",
      "862/862 [==============================] - 1s 986us/step - loss: 1.3945e-08 - acc: 0.7193 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 28/60\n",
      "862/862 [==============================] - 1s 982us/step - loss: 1.3945e-08 - acc: 0.7187 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 29/60\n",
      "862/862 [==============================] - 1s 999us/step - loss: 1.3945e-08 - acc: 0.7203 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 30/60\n",
      "862/862 [==============================] - 1s 964us/step - loss: 1.3945e-08 - acc: 0.7186 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 31/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7193 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 32/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7175 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 33/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7200 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 34/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7197 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 35/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7197 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 36/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7188 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 37/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7197 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 38/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7207 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 39/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7173 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 40/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7148 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 41/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7167 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 42/60\n",
      "862/862 [==============================] - 1s 2ms/step - loss: 1.3945e-08 - acc: 0.7212 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 43/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7187 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 44/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7196 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 45/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7191 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 46/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7193 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 47/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7185 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 48/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7194 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 49/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7193 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 50/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7193 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 51/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7206 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 52/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7190 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 53/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7198 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 54/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7214 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 55/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7176 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 56/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7187 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 57/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7174 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 58/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7168 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 59/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7177 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Epoch 60/60\n",
      "862/862 [==============================] - 1s 1ms/step - loss: 1.3945e-08 - acc: 0.7205 - val_loss: 1.3947e-08 - val_acc: 0.7176\n",
      "Wall time: 1min\n"
     ]
    }
   ],
   "source": [
    "# Fit the model\n",
    "%%time\n",
    "batch_size = input_dim\n",
    "epochs = 60\n",
    "\n",
    "model_history = model.fit(x_train, y_train,\n",
    "                    batch_size=batch_size,\n",
    "                    epochs=epochs,\n",
    "                    verbose=1,\n",
    "                    validation_data=(x_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "id": "olympic-resource",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 1.3947078159048942e-08\n",
      "Test accuracy: 0.7175716161727905\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmcAAAGQCAYAAADrzcctAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/d3fzzAAAACXBIWXMAAAsTAAALEwEAmpwYAAAkcUlEQVR4nO3debRddXn/8fdDiAQIEEyCQaZERJmEoGEoUAhgkaEKKihVwFKQH1WL0WLhp6Cx6k/4WStO1AJFlpVBgSJaERAZwmQhSIDIFIYgYUwCRCKE8ekfe184XO/JvffknnO+uXm/1jpr5+z93fs851nJvZ/sMTITSZIklWGlbhcgSZKk1xjOJEmSCmI4kyRJKojhTJIkqSCGM0mSpIIYziRJkgoyrMJZRJwREU9ExOwh2t5JETG7fn14EOttGhE3RMTzEXHMUsZ9KiLujYiMiHEN89eKiF9ExK0R8fuIOKxh2V4RcXe93nEN87euP/P2et01W/nOTer8bkQsHqrtSZKk5oZVOAPOBPYaig1FxL7AO4HJwPbA5/oKPBExt4/VnwSOBv6ln4+5Dng38GCv+Z8E7sjMrYGpwDcj4g0RMQL4PrA3sDnwNxGxeb3O6cBxmfkO4ELgc/189oBExBRgzFBsS5Ik9W9YhbPMnEEVjF4VERtHxCURcXNEXBMRmw5wc5sDV2fmS5n5J+BWBhj8MvOJzLwJeLGfcbdk5ty+FgFrREQAo6m+00vAdsC9mXl/Zr4AnAvsV6/zdmBG/edfAx8EiIgREfGNiLgpIm6LiP8zkO/Qsy7wDeCfBrqOJElaNsMqnDVxKvAPmfku4BjglAGudyuwd0SsVh9y3A3YoE019vY9YDPgEeB24NOZ+QqwHvBQw7h59TyA2cD76j8fyGu1Hg4sysxtgW2Bj0fEpAHW8Sng55n5aKtfRJIkDc7K3S6gnSJiNLAjcF61EwqAVeplHwD+uY/VHs7M92TmZRGxLXA9MB+4gWrvFRHxfWCnevybI2JW/efzMvNrQ1D6e4BZwO7AxsCvI+IaIPoY2/P8rb8DvhMRXwR+DrxQz98T2CoiDqjfrwVsEhF/oAqgfTkMeJgq5E1dpm8iSZIGZViHM6o9g09n5uTeCzLzv4D/WtrKddD6GkBEnA3Mqed/smdMRMzta/vL6DDgxKwefHpvRDwAbEq1p6xx7936VHvXyMy7qIIYEfE2YN+eEqn2HF7ax+ds2ayA+py7t9afD7BaRNybmW9dli8mSZKWblgf1szMPwIPRMSBAFHZeiDr1udqja3/vBWwFXBZ24p9vT8Ae9Sf/Saq88nuB26i2us1KSLeABxEtZeMiFinnq4EHA/8oN7WpcDfR8TIevnbImL1/grIzF9m5oTMnJiZE4FnDWaSJLXfsApnEXEO1eHHt0fEvIg4HPgocHhE3Ar8ntdOoO/PSOCaiLiD6ry1gzPzpQHWMSEi5gGfBY6va1mzXnZxRLy5/vPR9bj1gdsi4vR6E18BdoyI24HfAMdm5oL68z9FFbjuBH6amb+v1/mbiLgHuItqb9oP6/mnA3cAv6tvMfLvDP89ppIkLbeiOnImSZKkEhSz5ywi9o2Iy+q9TM9FxP0RcV5E/EW3a5MkSeqUIvacRcRJVPfSWgj8DFhAdTL6+6gOwR2amT/uWoGSJEkd0vVwFhETqG7bMB/YKjOfaFi2G3AF8EBmvmVp2xk3blxOnDixnaVKkiQNiZtvvnlBZo7va1kJJ4ZvRHV49X8agxlAZl4ZEc8AfRbfaOLEicycObNNJUqSJA2diOj96MZXlXDO2RyqG6Zu1/jwb4CI2AVYA7i8G4VJkiR1Wtf3nGXmkxFxLPCvwB0R8TOqc882pjrn7NfAgJ8HKUmStDzrejgDyMyTI2IucAbw8YZF9wJn9j7c2SMijgSOBNhwww3bXaYkSVLblXBYk4j4J+B84EyqPWarA++iuiv+WRHx//taLzNPzcwpmTll/Ph+T0uTJEkqXtfDWURMBU4Cfp6Zn83M+zPz2cz8HfB+qis5/zEilnq1piRJ0nDQ9XAG/HU9vbL3gsx8FriRqs5tOlmUJElSN5QQzlapp82OS/bMf6EDtUiSJHVVCeHsmnp6ZESs17ggIvYGdgKWANd3ujBJkqROK+FqzfOp7mP2buDOiLgQeAzYjOqQZwDHZebC7pUoSZLUGV0PZ5n5SkTsA3wSOIjqIoDVgCeBi4HvZOZlXSxRkiSpY7oezgAy80Xg5PolSZK0wioinC0Ppk2DWbO6XYUkSWq3yZPh5JO79/klXBAgSZKkmnvOBqibCVqSJK043HMmSZJUEMOZJElSQQxnkiRJBTGcSZIkFcRwJkmSVBDDmSRJUkEMZ5IkSQUxnEmSJBXEcCZJklQQw5kkSVJBDGeSJEkFMZxJkiQVxHAmSZJUEMOZJElSQQxnkiRJBTGcSZIkFcRwJkmSVBDDmSRJUkEMZ5IkSQUxnEmSJBXEcCZJklQQw5kkSVJBDGeSJEkFMZxJkiQVxHAmSZJUEMOZJElSQQxnkiRJBTGcSZIkFcRwJkmSVBDDmSRJUkEMZ5IkSQUxnEmSJBXEcCZJklQQw5kkSVJBDGeSJEkFMZxJkiQVxHAmSZJUEMOZJElSQQxnkiRJBTGcSZIkFcRwJkmSVBDDmSRJUkEMZ5IkSQUxnEmSJBXEcCZJklQQw5kkSVJBDGeSJEkFMZxJkiQVxHAmSZJUEMOZJElSQQxnkiRJBTGcSZIkFcRwJkmSVBDDmSRJUkEMZ5IkSQUxnEmSJBXEcCZJklQQw5kkSVJBDGeSJEkFMZxJkiQVxHAmSZJUEMOZJElSQYoKZxHxlxFxQUQ8GhHP19PLImKfbtcmSZLUCSt3u4AeEXE88BVgAfDfwKPAOGAbYCpwcdeKkyRJ6pAiwllEHEgVzC4HPpCZz/RaPrIrhUmSJHVY1w9rRsRKwEnAs8BHegczgMx8seOFSZIkdUEJe852BCYB5wNPRcS+wJbAEuDGzLyhm8VJkiR1UgnhbNt6+jjwO+AdjQsjYgZwQGbO73RhkiRJndb1w5rAOvX0KGBV4N3AGlR7zy4FdgHO62vFiDgyImZGxMz5881ukiRp+VdCOBtRT4NqD9lvMnNxZv4eeD8wD9g1Iv6i94qZeWpmTsnMKePHj+9gyZIkSe1RQjh7qp7en5m3Ni7IzOeo9p4BbNfRqiRJkrqghHB2dz19usnynvC2avtLkSRJ6q4SwtkM4CVgk4h4Qx/Lt6yncztWkSRJUpd0PZxl5gLgJ8BawBcbl0XEXwHvARYBl3S+OkmSpM4q4VYaAJ8Ftge+EBG7ADcCG1FdEPAy8PHMfLp75UmSJHVGEeEsM5+IiO2B46kC2Q7AM8Avga9n5m+7WZ8kSVKnFBHOADLzSao9aJ/tdi2SJEnd0vVzziRJkvQaw5kkSVJBDGeSJEkFMZxJkiQVxHAmSZJUEMOZJElSQQxnkiRJBTGcSZIkFcRwJkmSVBDDmSRJUkEMZ5IkSQUxnEmSJBXEcCZJklQQw5kkSVJBDGeSJEkFMZxJkiQVxHAmSZJUEMOZJElSQQxnkiRJBTGcSZIkFcRwJkmSVBDDmSRJUkEMZ5IkSQUxnEmSJBXEcCZJklQQw5kkSVJBDGeSJEkFMZxJkiQVxHAmSZJUEMOZJElSQQxnkiRJBTGcSZIkFcRwJkmSVBDDmSRJUkEMZ5IkSQUxnEmSJBXEcCZJklQQw5kkSVJBDGeSJEkFMZxJkiQVxHAmSZJUEMOZJElSQQxnkiRJBTGcSZIkFcRwJkmSVBDDmSRJUkEMZ5IkSQUxnEmSJBXEcCZJklQQw5kkSVJBDGeSJEkFMZxJkiQVZNDhLCLWjojNI2KVXvMPi4iLIuLsiNhu6EqUJElacazcwjr/DzgYWKdnRkT8A3AyEPWs/SNiSmbescwVSpKkYefFF19k3rx5LFmypNultNWoUaNYf/31GTly5IDXaSWc7QT8JjOfa5h3DPAw8BFgAvAj4LPAES1sX5IkDXPz5s1jjTXWYOLEiURE/ysshzKThQsXMm/ePCZNmjTg9Vo552w94IGeNxGxObAB8N3MvDYzzwd+AezSwrYlSdIKYMmSJYwdO3bYBjOAiGDs2LGD3jvYSjhbFWj8lJ2ABC5vmHcfVYiTJEnq03AOZj1a+Y6thLOHgU0b3r8H+CNwa8O8tYHGw56SJEkagFbC2ZXAPhHxqYg4AngfcElmvtIw5q3AQ0NRoCRJ0lB7+umnOeWUUwa93j777MPTTz899AU1aCWcfR1YDHwbOJXqEOf0noURsQ6wK3D9ENQnSZI05JqFs5dffnmp61188cWMGTOmTVVVBn21ZmY+EBFbAAfUs36emX9oGLIR8H3g7CGoT5Ikacgdd9xx3HfffUyePJmRI0cyevRo1l13XWbNmsUdd9zB/vvvz0MPPcSSJUv49Kc/zZFHHgnAxIkTmTlzJosXL2bvvfdm55135vrrr2e99dbjoosuYtVVV13m2lq5lQaZ+RjwvSbLbgJuWpaiJEnSCmTaNJg1a2i3OXkynHxy08Unnngis2fPZtasWVx11VXsu+++zJ49+9VbXpxxxhm88Y1v5LnnnmPbbbflgx/8IGPHjn3dNubMmcM555zDaaedxoc+9CEuuOACDj744GUuvaVw1peIGAf8JfAscHlmLn2/oCRJUiG22267192L7Dvf+Q4XXnghAA899BBz5sz5s3A2adIkJk+eDMC73vUu5s6dOyS1DDqcRcTfA38L7J2ZT9bz3gVcAryxHjYzInbPzD8NSZWSJGn4Wsoerk5ZffXVX/3zVVddxeWXX84NN9zAaqutxtSpU/u8V9kqq7z2JMsRI0bw3HNDc6OKVi4I+DCQPcGs9g2q22f8ELgY2BY4atnLkyRJGnprrLEGzzzzTJ/LFi1axNprr81qq63GXXfdxW9/+9uO1tZKONsEuK3nTX04c1fgPzLziMx8L9U5Zx9ptaiIOCQisn75CChJkjSkxo4dy0477cSWW27J5z73udct22uvvXjppZfYaqutOOGEE9hhhx06Wlsr55yNBZ5oeL9TPb2wYd41VIc+By0iNgC+S3W7jtGtbEOSJKk/Z5/d940lVlllFX71q1/1uaznvLJx48Yxe/bsV+cfc8wxQ1ZXK3vOngTGNbzfFXiF19/XLIFRg91wVM84+CGwEPhBC7VJkiQt11oJZ3cC742IsRExhuoctJsy848NYyYCj7Ww7aOB3YHDAC8mkCRJK5xWwtm3gXWBeVSPaJoAvHqL3YgYAezM65+12a+I2Aw4Efh2Zs5ooS5JkqTlXitPCPh5RBwFHFnPOiszf9ww5N1UhzQvHeg2I2Jl4D+BPwCfH2xNkiRJw0WrTwg4leq5mn0tu5TqthqD8UVgG2DnzBzwTUIi4kjqkLjhhhsO8iMlSZLK08phzSEVEdtR7S37ZmbeMJh1M/PUzJySmVPGjx/fngIlSZI6qOVwFhE7RMTpEXFzRNwXEb+LiNMiYsdBbKPncOY9wAmt1iJJktROo0d37u5eLR3WjIivAv8XiF6LJgN/FxEnZeZAzh0bDbyt/vOS6k4af+a0iDiN6kKBaa3UK0mStLxo5dmaB1IdhnwQ+ApwBfAo1RWcu1PtATs2ImZl5k/72dzzwH80WfZOqvPQrgXuBgZ1yFOSJKmZY489lo022ohPfOITAEyfPp2IYMaMGTz11FO8+OKLfPWrX2W//fbreG2RmYNbIWIG1SOc3pGZC/pYPg6YDdydmbu2XFjEdOBLwMcz8/T+xk+ZMiVnzpzZ6sdJkqQOuvPOO9lss80AmDYNZs0a2u1Pnrz056nfcsstTJs2jauvvhqAzTffnEsuuYQxY8aw5pprsmDBAnbYYQfmzJlDRDB69GgWL17cUi2N37VHRNycmVP6Gt/KYc2tgR/1FcwAMnNBRJwHHNrCtiVJktpum2224YknnuCRRx5h/vz5rL322qy77rp85jOfYcaMGay00ko8/PDDPP7440yYMKGjtbUSzlYGnu1nzLMtbluSJK1glraHq50OOOAAzj//fB577DEOOuggzjrrLObPn8/NN9/MyJEjmThxIkuWLOl4Xa1crXkv8NcR0ee69fx9gPuWpbDMnJ6ZMZBDmpIkSYN10EEHce6553L++edzwAEHsGjRItZZZx1GjhzJlVdeyYMPPtiVuloJZ+cAmwEXRcQmjQsiYmPgfGBzoO9HvUuSJBVgiy224JlnnmG99dZj3XXX5aMf/SgzZ85kypQpnHXWWWy66aZdqauVQ4//CuwF7AvsHRGPUF2tOQFYjyrwXVuPkyRJKtbtt9/+6p/HjRvHDTf0fXOIVi8GaMWg95xl5gvAXwFfAB4A1ge2BTao338B2KMeJ0mSpEFo9dmaLwJfB74eEaOBtYBFmbkYICJGRcSqmfnHoStVkiRp+FvmZ2tm5uLMfLgnmNX+DXhyWbctSZKGr8Hea3V51Mp3bOeDz/t8FpMkSdKoUaNYuHDhsA5omcnChQsZNWrUoNbzXmSSJKnj1l9/febNm8f8+fO7XUpbjRo1ivXXX39Q6xjOJElSx40cOZJJkyZ1u4witfOwpiRJkgbJcCZJklQQw5kkSVJBBnTOWUS83O5CJEmSNPALAlq5LcbwvTZWkiSpTQYUzjLTw5+SJEkdYOiSJEkqiOFMkiSpIIYzSZKkghjOJEmSCmI4kyRJKojhTJIkqSCGM0mSpIIYziRJkgpiOJMkSSqI4UySJKkghjNJkqSCGM4kSZIKYjiTJEkqiOFMkiSpIIYzSZKkghjOJEmSCmI4kyRJKojhTJIkqSCGM0mSpIIYziRJkgpiOJMkSSqI4UySJKkghjNJkqSCGM4kSZIKYjiTJEkqiOFMkiSpIIYzSZKkghjOJEmSCmI4kyRJKojhTJIkqSCGM0mSpIIYziRJkgpiOJMkSSqI4UySJKkghjNJkqSCGM4kSZIKYjiTJEkqiOFMkiSpIIYzSZKkghjOJEmSCmI4kyRJKojhTJIkqSCGM0mSpIIYziRJkgpiOJMkSSqI4UySJKkghjNJkqSCGM4kSZIKYjiTJEkqiOFMkiSpIIYzSZKkgnQ9nEXE2Ig4IiIujIh7I+K5iFgUEddGxOER0fUaJUmSOmXlbhcAHAj8G/AocCXwB+BNwAeA04G9I+LAzMzulShJktQZJYSze4D3Ab/MzFd6ZkbE54EbgQ9SBbULulOeJElS53T9kGFmXpGZv2gMZvX8x4Af1G+ndrwwSZKkLuh6OOvHi/X0pa5WIUmS1CHFhrOIWBk4tH57STdrkSRJ6pRiwxlwIrAlcHFmXtrXgIg4MiJmRsTM+fPnd7Y6SZKkNigynEXE0cA/AncBhzQbl5mnZuaUzJwyfvz4jtUnSZLULsWFs4j4JPBt4A5gt8x8ssslSZIkdUxR4SwipgHfA2ZTBbPHuluRJElSZxUTziLiWOBbwCyqYPZEdyuSJEnqvCLCWUScQHUBwM3AHpm5oMslSZIkdUXXnxAQER8D/hl4GbgGODoieg+bm5lndrg0SZKkjut6OAMm1dMRwLQmY64GzuxEMZIkSd3U9cOamTk9M6Of19Ru1ylJktQJXQ9nkiRJeo3hTJIkqSCGM0mSpIIYziRJkgpiOJMkSSqI4UySJKkghjNJkqSCGM4kSZIKYjiTJEkqiOFMkiSpIIYzSZKkghjOJEmSCmI4kyRJKojhTJIkqSCGM0mSpIIYziRJkgpiOJMkSSqI4UySJKkghjNJkqSCGM4kSZIKYjiTJEkqiOFMkiSpIIYzSZKkghjOJEmSCmI4kyRJKojhTJIkqSCGM0mSpIIYziRJkgpiOJMkSSqI4UySJKkghjNJkqSCGM4kSZIKYjiTJEkqiOFMkiSpIIYzSZKkghjOJEmSCmI4kyRJKojhTJIkqSCGM0mSpIIYziRJkgpiOJMkSSqI4UySJKkghjNJkqSCGM4kSZIKYjiTJEkqiOFMkiSpIIYzSZKkghjOJEmSCmI4kyRJKojhTJIkqSCGM0mSpIIYziRJkgpiOJMkSSqI4UySJKkghjNJkqSCGM4kSZIKYjiTJEkqyMrdLmC5MW0azJrV7SokSVK7TZ4MJ5/ctY93z5kkSVJB3HM2UF1M0JIkacXhnjNJkqSCGM4kSZIKYjiTJEkqiOFMkiSpIIYzSZKkghQTziJi/Yg4IyIeiYjnI2JuRJwcEWt3uzZJkqROKeJWGhGxMXA9sA5wEXAXsB3waWCviNgpMxd2sURJkqSOKGXP2SlUwezozNw/M4/LzN2BbwFvB77W1eokSZI6pOvhLCLeAuwJzAW+32vxl4A/AYdExOodLk2SJKnjuh7OgN3r6WWZ+Urjgsx8BrgOWA3YodOFSZIkdVoJ4ezt9fSeJsvn1NO3daAWSZKkriohnK1VTxc1Wd4zf0zvBRFxZETMjIiZ8+fPb0dtkiRJHVVCOOtP1NPsvSAzT83MKZk5Zfz48R0uS5IkaeiVcCuNnj1jazVZvmavcX26+eabF0TEg0NWVd/GAQva/BkrIvvaHva1Pexre9jX9rCv7TEUfd2o2YISwtnd9bTZOWWb1NNm56QBkJlt33UWETMzc0q7P2dFY1/bw762h31tD/vaHva1Pdrd1xIOa15ZT/eMiNfVExFrADsBzwG/7XRhkiRJndb1cJaZ9wGXAROBT/Za/GVgdeBHmfmnDpcmSZLUcSUc1gT4BNXjm74TEXsAdwLbA7tRHc78Qhdra3RqtwsYpuxre9jX9rCv7WFf28O+tkdb+xqZf3YRZFdExAbAPwN7AWOBR4GfAV/OzCe7WJokSVLHFBPOJEmSVMA5Z5IkSXqN4awfEbF+RJwREY9ExPMRMTciTo6ItbtdW+ki4oCI+G5EXBMRf4yIjIgf97POjhFxcUQ8GRHPRsRtETEtIkZ0qu6SRcTYiDgiIi6MiHsj4rmIWBQR10bE4b2veG5Yz772IyJOiojfRMRDdV+fjIhbIuJLETG2yTr2dZAi4pD6Z0FGxBFNxtjXftS/i7LJ67Em69jXAYqIv4yICyLi0fp3/6MRcVlE7NPH2CHvq4c1lyIiNqa6UGEd4CLgLmA7qgsV7gZ2ysyF3auwbBExC9gaWAzMAzYFzsrMg5uM3w+4AFgC/AR4Engv1fNXz8/MAztQdtEi4ijg36jOybwS+APwJuADVDdyvgA4MBv+YdvXgYmIF4DfAXcAT1BdKb4DMAV4BNghMx9qGG9fB6k+t/h2YAQwGvh4Zp7ea4x9HYCImEv1WMOT+1i8ODP/pdd4+zpAEXE88BWqm8z+N9XP23HANsCVmflPDWPb09fM9NXkBVxK9diof+g1/1/r+T/odo0lv6hC7CZUj+CaWvfsx03Grkn1C/F5YErD/FFUATmBg7r9nbr9Anav/+Gv1Gv+BKqglsAH7WtLvR3VZP7X6j6dYl+Xqb8BXA7cB3yj7tERvcbY14H3cy4wd4Bj7evA+3pg3Y9fA2v0sXxkJ/rqYc0mIuItwJ5U/wC+32vxl4A/AYdExOodLm25kZlXZuacrP+29uMAYDxwbmbObNjGEuD4+u3ft6HM5UpmXpGZv8jMV3rNfwz4Qf12asMi+zpAdU/68tN6uknDPPs6eEdT/efiMKqfn32xr+1hXwegPi3kJOBZ4COZ+UzvMZn5YsPbtvW1lPuclWj3enpZH78In4mI66jC2w7Abzpd3DDU0+9L+lg2g+ofy44RsUpmPt+5spYrPT80XmqYZ1+X3Xvr6W0N8+zrIETEZsCJwLczc0ZE7N5kqH0dnFUi4mBgQ6rAexswIzNf7jXOvg7MjsAk4HzgqYjYF9iS6pDljZl5Q6/xbeur4ay5t9fTZs/0nEMVzt6G4WwoNO13Zr4UEQ8AWwBvobpJsRpExMrAofXbxh8U9nWQIuIYqvOh1qI632xnql96JzYMs68DVP/d/E+qw+6f72e4fR2cCVS9bfRARByWmVc3zLOvA7NtPX2c6vzTdzQujIgZwAGZOb+e1ba+elizubXq6aImy3vmj2l/KSsE+71sTqT6H97FmXlpw3z7OnjHUJ26MI0qmF0C7NnwAxns62B8kepE6r/NzOf6GWtfB+6HwB5UAW11qiDx71SPQvxVRGzdMNa+Dsw69fQoYFXg3cAaVD9bLwV2Ac5rGN+2vhrOWhf11MtdO8N+NxERRwP/SHU18SGDXb2e2tdaZk7IzKD6pfcBqv/13hIR7xzEZuwrEBHbUe0t+2Yfh4Ra2mQ9XaH7CpCZX67PQX08M5/NzNmZeRTVBWurAtMHsTn7Wum59UVQ7SH7TWYuzszfA++nuuvArhHxFwPcXst9NZw115N412qyfM1e47Rs7HcLIuKTwLepbv+wW/75o87sa4vqX3oXUp2+MBb4UcNi+9qPhsOZ9wAnDHA1+7rsei4M2qVhnn0dmKfq6f2ZeWvjgnqvb89Rie3qadv6ajhr7u56+rYmy3uu3Gp2TpoGp2m/6x/yk6hOdL+/k0WVLCKmAd8DZlMFs75uPGlfl1FmPkgVfreIiHH1bPvav9FU/dkMWNJ4k1Sqw8YAp9XzTq7f29dl90Q9bbyTgH0dmJ4+Pd1keU94W7XX+CHvq+GsuSvr6Z6977oeEWsAOwHPAb/tdGHD1BX1dK8+lu0CrAZcv4JfSfSqiDgW+BYwiyqYPdFkqH0dGm+upz1XwdnX/j0P/EeT1y31mGvr9z2HPO3rsus55NYYCOzrwMygClObRMQb+li+ZT2dW0/b19du3/Ct5BfehHYoezmV/m9COx9vkjiQXp5Q92Mm8MZ+xtrXgfV0U2BCH/NX4rWb0F5nX4es39NpfhNa+9p//7bo698+sBHVnQQS+Lx9bam3P6778dVe8/8KeIVqr9qYdvfVxzctRR+Pb7oT2J7qzvf3ADumj29qKiL2B/av304A3kP1v7lr6nkLMvOYXuPPp7qnzLlUj8F4H/VjMIAP5Qr+FzYiPgacSbUH57v0fS7D3Mw8s2Gd/bGvS1UfIv4G1f+c7wMWUj0Wa1eqCwIeA/bIzDsa1tkf+9qSiJhOdWizr8c37Y99Xaq6f8dRHeF5AHgG2BjYlyoYXAy8PzNfaFhnf+xrvyJiHeA64K1Uv6tupAq976cKWx/JzPMaxu9PO/ra7ZRa+gvYgOqS5UeBF4AHqU7AXuoeC1+v+99xs9fcPtbZieoHy1NUh41vBz4DjOj29ynhNYCeJnCVfR10X7ekehLILKrn6b1EFXxvqnve5793+9pyv3v+Hh/RZLl9XXr/dgXOobpC+2mqG1DPp3rk0KHUz822ry33941UR8geqH/vL6TaQbNDp/rqnjNJkqSCeEGAJElSQQxnkiRJBTGcSZIkFcRwJkmSVBDDmSRJUkEMZ5IkSQUxnEmSJBXEcCZJHRIR0+sHfU/tdi2SymU4k7TcqINNf6+p3a5TkpbFyt0uQJJa8OWlLJvbqSIkqR0MZ5KWO5k5vds1SFK7eFhT0rDVeI5XRHwsIm6JiOci4omIOCMiJjRZb5OI+FFEPBwRL0TEI/X7TZqMHxERR0XEdRGxqP6MeyPi9KWsc0BE3BgRz0bEkxFxbkSsN5TfX9LyyT1nklYEnwH2BH4CXALsDBwGTI2I7TNzfs/AiNgWuBxYA/g5cAewKfBRYL+I2CMzZzaMfwPwS+DdwEPA2cAfgYnA+4FrgTm96vkE8L56+1cD2wMfBraOiMmZ+fxQfnlJyxfDmaTlTkRMb7JoSWae2Mf8vYHtM/OWhm18C5gGnAgcXs8L4EfAmsDBmXlWw/gPA+cCP46IzTPzlXrRdKpg9gvgwMZgFRGr1NvqbS9g28y8vWHs2cDfAPsBP2323SUNf5GZ3a5BkgYkIvr7gbUoM8c0jJ8OfAk4IzMP77WttYAHgVWAMZn5fETsRLWn64bM3LGPz7+Gaq/brpk5IyJGAAuBNwBvzcxH+qm/p56vZebxvZbtBlwBfDMzj+nne0oaxjznTNJyJzOjyWtMk1Wu7mMbi4BZwChgs3r2O+vpFU220zN/m3q6KbAWcFt/wayXmX3Me6ierj2I7UgahgxnklYEjzeZ/1g9XavX9NEm43vmj+k1fXiQ9Tzdx7yX6umIQW5L0jBjOJO0InhTk/k9V2su6jXt8ypOYN1e456up15lKWnIGM4krQh27T2jPudsMrAEuLOe3XPBwNQm2+mZ/7t6ehdVQNsqIt687GVKkuFM0orhkIjYpte86VSHMc9puMLyOuBuYOeIOKBxcP1+F+AeqosGyMyXgVOAVYEf1FdnNq7zhogYP8TfRdIw5600JC13lnIrDYCfZeasXvN+BVwXET+lOm9s5/o1FziuZ1BmZkR8DPg18JOIuIhq79jbgf2BZ4BDG26jAdWjpLYH3gvcExH/XY/bgOreap8Dzmzha0paQRnOJC2PvrSUZXOprsJs9C3gQqr7mn0YWEwVmD6fmU80DszM/6lvRHs81f3L3gssAM4BvpKZd/ca/0JE7AUcBRwKfAwI4JH6M68d7JeTtGLzPmeShq2G+4rtlplXdbcaSRoYzzmTJEkqiOFMkiSpIIYzSZKkgnjOmSRJUkHccyZJklQQw5kkSVJBDGeSJEkFMZxJkiQVxHAmSZJUEMOZJElSQf4X6ki3weFsPcgAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "score = model.evaluate(x_test, y_test, verbose=0)\n",
    "print('Test loss:', score[0])\n",
    "print('Test accuracy:', score[1])\n",
    "\n",
    "fig, ax = plt.subplots(1, 1, figsize=(10,6))\n",
    "ax.plot(np.sqrt(model_history.history['acc']), 'r', label='train_acc')\n",
    "ax.plot(np.sqrt(model_history.history['val_acc']), 'b' ,label='val_acc')\n",
    "ax.set_xlabel(r'Epoch', fontsize=20)\n",
    "ax.set_ylabel(r'Accuracy', fontsize=20)\n",
    "ax.legend()\n",
    "ax.tick_params(labelsize=20)\n",
    "\n",
    "fig, ax = plt.subplots(1, 1, figsize=(10,6))\n",
    "ax.plot(np.sqrt(model_history.history['loss']), 'r', label='train')\n",
    "ax.plot(np.sqrt(model_history.history['val_loss']), 'b' ,label='val')\n",
    "ax.set_xlabel(r'Epoch', fontsize=20)\n",
    "ax.set_ylabel(r'Loss', fontsize=20)\n",
    "ax.legend()\n",
    "ax.tick_params(labelsize=20)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "loving-system",
   "metadata": {},
   "source": [
    "### Conclusion"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "analyzed-pension",
   "metadata": {},
   "source": [
    "Unfortunately, for this data xgboost, ANN and DNN gave accuracy around 91% and also for the test data 89%. My intention was, how to tune the parameters for xgboost and ANN."
   ]
  }
 ],
 "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": 5
}