2. UNSW-NB15 - Parameter Tuning.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "7903d768",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Import libraries:\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import xgboost as xgb\n",
    "from xgboost.sklearn import XGBClassifier\n",
    "from sklearn import metrics   #Additional scklearn functions\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import GridSearchCV   #Perforing grid search\n",
    "\n",
    "import matplotlib.pylab as plt\n",
    "%matplotlib inline\n",
    "from matplotlib.pylab import rcParams\n",
    "rcParams['figure.figsize'] = 12, 4\n",
    "\n",
    "train = pd.read_csv('UNSW_NB15_bayesian_enc_training_set.csv')\n",
    "test = pd.read_csv('UNSW_NB15_bayesian_enc_testing_set.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "0f40825c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    119341\n",
       "0     56000\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target = 'label'\n",
    "IDcol = 'id'\n",
    "ACcol = 'attack_cat'\n",
    "train['label'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "b5e2f27c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    45332\n",
       "0    37000\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test['label'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "0e6a5f62",
   "metadata": {},
   "outputs": [],
   "source": [
    "def modelfit(alg, dtrain, dtest, predictors,useTrainCV=True, cv_folds=5, early_stopping_rounds=50):\n",
    "    \n",
    "    if useTrainCV:\n",
    "        xgb_param = alg.get_xgb_params()\n",
    "        xgtrain = xgb.DMatrix(dtrain[predictors].values, label=dtrain[target].values)\n",
    "        xgtest = xgb.DMatrix(dtest[predictors].values)\n",
    "        cvresult = xgb.cv(xgb_param, xgtrain, num_boost_round=alg.get_params()['n_estimators'], nfold=cv_folds,\n",
    "            metrics='auc', early_stopping_rounds=early_stopping_rounds)\n",
    "        n_estimators=cvresult.shape[0]\n",
    "        alg.set_params(n_estimators=cvresult.shape[0])\n",
    "        print(cvresult)\n",
    "        print(n_estimators)\n",
    "    \n",
    "    #Fit the algorithm on the data\n",
    "    alg.fit(dtrain[predictors], dtrain['label'],eval_metric='auc')\n",
    "        \n",
    "    #Predict training set:\n",
    "    dtrain_predictions = alg.predict(dtrain[predictors])\n",
    "    dtrain_predprob = alg.predict_proba(dtrain[predictors])[:,1]\n",
    "    print(dtrain_predictions)\n",
    "    print(dtrain_predprob )\n",
    "        \n",
    "    #Print model report:\n",
    "    print (\"\\nModel Report\")\n",
    "    print (\"Accuracy : %.4g\" % metrics.accuracy_score(dtrain['label'].values, dtrain_predictions))\n",
    "    print (\"AUC Score (Train): %f\" % metrics.roc_auc_score(dtrain['label'], dtrain_predprob))\n",
    "    \n",
    "#     Predict on testing data:\n",
    "    dtest_predprob = alg.predict_proba(dtest[predictors])[:,1]\n",
    "    print ('AUC Score (Test): %f' % metrics.roc_auc_score(dtest['label'], dtest_predprob))\n",
    "                \n",
    "    feat_imp = pd.Series(alg.get_booster().get_fscore()).sort_values(ascending=False)\n",
    "    feat_imp.plot(kind='bar', title='Feature Importances')\n",
    "    plt.ylabel('Feature Importance Score')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "dc57dbab",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "      train-auc-mean  train-auc-std  test-auc-mean  test-auc-std\n",
      "0           0.954683       0.006009       0.954162      0.005136\n",
      "1           0.974192       0.004496       0.973776      0.005485\n",
      "2           0.979690       0.002150       0.979447      0.002692\n",
      "3           0.981730       0.001325       0.981518      0.002134\n",
      "4           0.982909       0.001122       0.982751      0.001769\n",
      "...              ...            ...            ...           ...\n",
      "1479        0.996959       0.000135       0.993515      0.000336\n",
      "1480        0.996959       0.000136       0.993515      0.000337\n",
      "1481        0.996961       0.000136       0.993515      0.000337\n",
      "1482        0.996962       0.000136       0.993516      0.000334\n",
      "1483        0.996963       0.000137       0.993516      0.000334\n",
      "\n",
      "[1484 rows x 4 columns]\n",
      "1484\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\admin\\anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1 0 1 ... 1 1 1]\n",
      "[0.56998926 0.03172128 0.7467671  ... 0.9998087  0.99999225 0.99999225]\n",
      "\n",
      "Model Report\n",
      "Accuracy : 0.9723\n",
      "AUC Score (Train): 0.996902\n",
      "AUC Score (Test): 0.982351\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "predictors = [x for x in train.columns if x not in [target, IDcol, ACcol]]\n",
    "xgb1 = XGBClassifier(\n",
    "        learning_rate =0.1,\n",
    "        n_estimators=1484,\n",
    "        max_depth=4,\n",
    "        min_child_weight=12,\n",
    "        gamma=0.3,\n",
    "        alpha = 0.05,\n",
    "        subsample=0.6,\n",
    "        colsample_bytree=0.6,\n",
    "        objective= 'binary:logistic',\n",
    "        nthread=4,\n",
    "        scale_pos_weight=1,\n",
    "        seed=27)\n",
    "modelfit(xgb1, train, test, predictors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "34b3a853",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\admin\\anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[09:39:16] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "({'mean_fit_time': array([ 729.07954907,  721.28482447,  705.19047151, 1174.94892302,\n",
       "         1136.23560157, 1093.03463545, 1663.33019209, 1565.91993876,\n",
       "         1505.46529331, 2136.15486083, 1992.03475027, 1737.89494276]),\n",
       "  'std_fit_time': array([  2.45632772,   6.09019432,   9.64368213,  28.52239646,\n",
       "           7.20569349,  31.90376791,  48.27968294,  18.33054567,\n",
       "          62.32298757,  66.45935853,  31.40903026, 227.89799608]),\n",
       "  'mean_score_time': array([0.67495375, 0.71870208, 0.66245537, 1.07492681, 1.10304961,\n",
       "         1.10929971, 1.58114166, 1.50927229, 1.65926151, 1.87799745,\n",
       "         1.91549435, 1.7361239 ]),\n",
       "  'std_score_time': array([0.05537125, 0.08148442, 0.03775702, 0.06874522, 0.02724108,\n",
       "         0.04940706, 0.10976865, 0.05978014, 0.21063835, 0.12475602,\n",
       "         0.0830285 , 0.26817182]),\n",
       "  'param_max_depth': masked_array(data=[3, 3, 3, 5, 5, 5, 7, 7, 7, 9, 9, 9],\n",
       "               mask=[False, False, False, False, False, False, False, False,\n",
       "                     False, False, False, False],\n",
       "         fill_value='?',\n",
       "              dtype=object),\n",
       "  'param_min_child_weight': masked_array(data=[1, 3, 5, 1, 3, 5, 1, 3, 5, 1, 3, 5],\n",
       "               mask=[False, False, False, False, False, False, False, False,\n",
       "                     False, False, False, False],\n",
       "         fill_value='?',\n",
       "              dtype=object),\n",
       "  'params': [{'max_depth': 3, 'min_child_weight': 1},\n",
       "   {'max_depth': 3, 'min_child_weight': 3},\n",
       "   {'max_depth': 3, 'min_child_weight': 5},\n",
       "   {'max_depth': 5, 'min_child_weight': 1},\n",
       "   {'max_depth': 5, 'min_child_weight': 3},\n",
       "   {'max_depth': 5, 'min_child_weight': 5},\n",
       "   {'max_depth': 7, 'min_child_weight': 1},\n",
       "   {'max_depth': 7, 'min_child_weight': 3},\n",
       "   {'max_depth': 7, 'min_child_weight': 5},\n",
       "   {'max_depth': 9, 'min_child_weight': 1},\n",
       "   {'max_depth': 9, 'min_child_weight': 3},\n",
       "   {'max_depth': 9, 'min_child_weight': 5}],\n",
       "  'split0_test_score': array([0.99554244, 0.99508918, 0.99514837, 0.99527759, 0.99535158,\n",
       "         0.99541283, 0.99536256, 0.995325  , 0.99546396, 0.9954506 ,\n",
       "         0.99544213, 0.99533251]),\n",
       "  'split1_test_score': array([0.99847559, 0.9985761 , 0.99847211, 0.99854981, 0.99859067,\n",
       "         0.99855776, 0.99847428, 0.99849257, 0.99855198, 0.99854248,\n",
       "         0.99854064, 0.99849493]),\n",
       "  'split2_test_score': array([0.99925066, 0.99923841, 0.99924094, 0.99921311, 0.99924044,\n",
       "         0.99920883, 0.99919851, 0.99916375, 0.99916702, 0.99916903,\n",
       "         0.99916858, 0.99917347]),\n",
       "  'split3_test_score': array([0.99261544, 0.99251946, 0.99249558, 0.99187965, 0.9917824 ,\n",
       "         0.9918384 , 0.9911984 , 0.99129559, 0.99131718, 0.99047615,\n",
       "         0.99065427, 0.99081135]),\n",
       "  'split4_test_score': array([0.98078173, 0.98050364, 0.98064568, 0.9779908 , 0.97743334,\n",
       "         0.97851656, 0.97928449, 0.97856511, 0.97781686, 0.97841752,\n",
       "         0.97786965, 0.97793123]),\n",
       "  'mean_test_score': array([0.99333317, 0.99318536, 0.99320054, 0.99258219, 0.99247969,\n",
       "         0.99270688, 0.99270365, 0.9925684 , 0.9924634 , 0.99241115,\n",
       "         0.99233505, 0.9923487 ]),\n",
       "  'std_test_score': array([0.00669919, 0.00679107, 0.00672594, 0.00774918, 0.0079772 ,\n",
       "         0.0075624 , 0.00727864, 0.00753532, 0.00783387, 0.00764286,\n",
       "         0.00783496, 0.00779081]),\n",
       "  'rank_test_score': array([ 1,  3,  2,  6,  8,  4,  5,  7,  9, 10, 12, 11])},\n",
       " {'max_depth': 3, 'min_child_weight': 1},\n",
       " 0.9933331724220043)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# min_child_weight and max_depth\n",
    "param_test1 = {\n",
    " 'max_depth':range(3,10,2),\n",
    " 'min_child_weight':range(1,6,2)\n",
    "}\n",
    "gsearch1 = GridSearchCV(estimator = XGBClassifier( learning_rate =0.1, n_estimators=1484, max_depth=5,\n",
    " min_child_weight=1, gamma=0, subsample=0.8, colsample_bytree=0.8,\n",
    " objective= 'binary:logistic', nthread=4, scale_pos_weight=1, seed=27), \n",
    " param_grid = param_test1, scoring='roc_auc',n_jobs=4, cv=5)\n",
    "gsearch1.fit(train[predictors],train[target])\n",
    "gsearch1.cv_results_, gsearch1.best_params_, gsearch1.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "fd5bc9be",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\admin\\anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[13:40:23] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "({'mean_fit_time': array([ 952.62244406,  931.34406414,  993.40586224, 1255.31639724,\n",
       "         1292.63907628, 1224.4488276 , 1550.41826186, 1476.49587364,\n",
       "         1260.52944708]),\n",
       "  'std_fit_time': array([ 13.40141049,  13.16698588,  44.72532571,  31.06582183,\n",
       "          30.95756219,  12.04244689,  25.51923375,  21.44286864,\n",
       "         322.663164  ]),\n",
       "  'mean_score_time': array([1.00930686, 1.09055128, 1.20190148, 1.28014998, 3.77934427,\n",
       "         1.25090842, 9.42779179, 1.79039021, 1.38934655]),\n",
       "  'std_score_time': array([0.08534727, 0.19272686, 0.2683603 , 0.24837499, 2.07052474,\n",
       "         0.05440438, 7.67327306, 0.39540293, 0.29376309]),\n",
       "  'param_max_depth': masked_array(data=[4, 4, 4, 5, 5, 5, 6, 6, 6],\n",
       "               mask=[False, False, False, False, False, False, False, False,\n",
       "                     False],\n",
       "         fill_value='?',\n",
       "              dtype=object),\n",
       "  'param_min_child_weight': masked_array(data=[4, 5, 6, 4, 5, 6, 4, 5, 6],\n",
       "               mask=[False, False, False, False, False, False, False, False,\n",
       "                     False],\n",
       "         fill_value='?',\n",
       "              dtype=object),\n",
       "  'params': [{'max_depth': 4, 'min_child_weight': 4},\n",
       "   {'max_depth': 4, 'min_child_weight': 5},\n",
       "   {'max_depth': 4, 'min_child_weight': 6},\n",
       "   {'max_depth': 5, 'min_child_weight': 4},\n",
       "   {'max_depth': 5, 'min_child_weight': 5},\n",
       "   {'max_depth': 5, 'min_child_weight': 6},\n",
       "   {'max_depth': 6, 'min_child_weight': 4},\n",
       "   {'max_depth': 6, 'min_child_weight': 5},\n",
       "   {'max_depth': 6, 'min_child_weight': 6}],\n",
       "  'split0_test_score': array([0.99527746, 0.99540918, 0.99526899, 0.99529308, 0.99541283,\n",
       "         0.99549453, 0.99544469, 0.99577775, 0.99557315]),\n",
       "  'split1_test_score': array([0.99858037, 0.99852784, 0.99858076, 0.99854772, 0.99855776,\n",
       "         0.99855145, 0.99849639, 0.99858853, 0.99853959]),\n",
       "  'split2_test_score': array([0.9992016 , 0.99923612, 0.9992491 , 0.99917695, 0.99920883,\n",
       "         0.99919499, 0.99919234, 0.99920065, 0.99915955]),\n",
       "  'split3_test_score': array([0.99214159, 0.99207326, 0.9921414 , 0.99196124, 0.9918384 ,\n",
       "         0.99184588, 0.99149526, 0.9915091 , 0.99150372]),\n",
       "  'split4_test_score': array([0.97956054, 0.98024202, 0.97937974, 0.97818197, 0.97851656,\n",
       "         0.97821732, 0.97815327, 0.97785551, 0.97778082]),\n",
       "  'mean_test_score': array([0.99295231, 0.99309768, 0.992924  , 0.99263219, 0.99270688,\n",
       "         0.99266083, 0.99255639, 0.99258631, 0.99251137]),\n",
       "  'std_test_score': array([0.00715726, 0.00691126, 0.00723276, 0.00767029, 0.0075624 ,\n",
       "         0.00767729, 0.00769724, 0.00785007, 0.0078475 ]),\n",
       "  'rank_test_score': array([2, 1, 3, 6, 4, 5, 8, 7, 9])},\n",
       " {'max_depth': 4, 'min_child_weight': 5},\n",
       " 0.9930976819061339)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# min_child_weight and max_depth\n",
    "param_test2 = {\n",
    " 'max_depth':[4,5,6],\n",
    " 'min_child_weight':[4,5,6]\n",
    "}\n",
    "gsearch2 = GridSearchCV(estimator = XGBClassifier( learning_rate=0.1, n_estimators=1484, max_depth=3,\n",
    " min_child_weight=1, gamma=0, subsample=0.8, colsample_bytree=0.8,\n",
    " objective= 'binary:logistic', nthread=4, scale_pos_weight=1,seed=27), \n",
    " param_grid = param_test2, scoring='roc_auc',n_jobs=4, cv=5)\n",
    "gsearch2.fit(train[predictors],train[target])\n",
    "gsearch2.cv_results_, gsearch2.best_params_, gsearch2.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "8fd9a20c",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\admin\\anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[20:03:43] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "({'mean_fit_time': array([1051.74756641,  946.71739564,  943.99094653,  925.68113279]),\n",
       "  'std_fit_time': array([57.41024096, 18.3816269 , 13.36508591,  3.79335233]),\n",
       "  'mean_score_time': array([6.00066304, 1.04559722, 0.95774961, 0.81208286]),\n",
       "  'std_score_time': array([2.77842274, 0.03872624, 0.05874464, 0.14548765]),\n",
       "  'param_min_child_weight': masked_array(data=[6, 8, 10, 12],\n",
       "               mask=[False, False, False, False],\n",
       "         fill_value='?',\n",
       "              dtype=object),\n",
       "  'params': [{'min_child_weight': 6},\n",
       "   {'min_child_weight': 8},\n",
       "   {'min_child_weight': 10},\n",
       "   {'min_child_weight': 12}],\n",
       "  'split0_test_score': array([0.99526899, 0.99521727, 0.99532411, 0.99543151]),\n",
       "  'split1_test_score': array([0.99858076, 0.99852888, 0.99858611, 0.99858531]),\n",
       "  'split2_test_score': array([0.9992491 , 0.9992114 , 0.99921756, 0.99920916]),\n",
       "  'split3_test_score': array([0.9921414 , 0.99219999, 0.99207334, 0.9922621 ]),\n",
       "  'split4_test_score': array([0.97937974, 0.97956487, 0.97956485, 0.97943044]),\n",
       "  'mean_test_score': array([0.992924  , 0.99294448, 0.99295319, 0.9929837 ]),\n",
       "  'std_test_score': array([0.00723276, 0.00714414, 0.00716401, 0.00721579]),\n",
       "  'rank_test_score': array([4, 3, 2, 1])},\n",
       " {'min_child_weight': 12},\n",
       " 0.9929837043972934)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# min_child_weight\n",
    "param_test2b = {'min_child_weight':[6,8,10,12]}\n",
    "gsearch2b = GridSearchCV(estimator = XGBClassifier( learning_rate=0.1, n_estimators=1484, max_depth=4,\n",
    " min_child_weight=2, gamma=0, subsample=0.8, colsample_bytree=0.8,\n",
    " objective= 'binary:logistic', nthread=4, scale_pos_weight=1,seed=27), \n",
    " param_grid = param_test2b, scoring='roc_auc',n_jobs=4, cv=5)\n",
    "gsearch2b.fit(train[predictors],train[target])\n",
    "gsearch2b.cv_results_, gsearch2b.best_params_, gsearch2b.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "86cf3d96",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\admin\\anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[22:30:48] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "({'mean_fit_time': array([1352.5876718 , 1325.01373353, 1331.15502725, 1327.53921032,\n",
       "         1137.33433867]),\n",
       "  'std_fit_time': array([ 32.56213844,  12.6507868 ,  13.92604445,  13.03730149,\n",
       "         362.38219525]),\n",
       "  'mean_score_time': array([1.35761366, 1.39270434, 1.37685523, 1.37685852, 1.1583261 ]),\n",
       "  'std_score_time': array([0.04465967, 0.02915961, 0.0438644 , 0.06813015, 0.1853099 ]),\n",
       "  'param_gamma': masked_array(data=[0.0, 0.1, 0.2, 0.3, 0.4],\n",
       "               mask=[False, False, False, False, False],\n",
       "         fill_value='?',\n",
       "              dtype=object),\n",
       "  'params': [{'gamma': 0.0},\n",
       "   {'gamma': 0.1},\n",
       "   {'gamma': 0.2},\n",
       "   {'gamma': 0.3},\n",
       "   {'gamma': 0.4}],\n",
       "  'split0_test_score': array([0.995556  , 0.99561346, 0.9954808 , 0.99545661, 0.99547685]),\n",
       "  'split1_test_score': array([0.99855378, 0.99853726, 0.9985147 , 0.99856914, 0.99860776]),\n",
       "  'split2_test_score': array([0.99913757, 0.99914831, 0.99920267, 0.99916078, 0.99917513]),\n",
       "  'split3_test_score': array([0.99168043, 0.99157151, 0.99149873, 0.99156542, 0.99156905]),\n",
       "  'split4_test_score': array([0.97842383, 0.97847199, 0.9780809 , 0.97785939, 0.97875852]),\n",
       "  'mean_test_score': array([0.99267032, 0.99266851, 0.99255556, 0.99252227, 0.99271746]),\n",
       "  'std_test_score': array([0.00759882, 0.00758745, 0.00773156, 0.00781231, 0.00748575]),\n",
       "  'rank_test_score': array([2, 3, 4, 5, 1])},\n",
       " {'gamma': 0.4},\n",
       " 0.992717460286984)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# gamma\n",
    "param_test3 = {\n",
    " 'gamma':[i/10.0 for i in range(0,5)]\n",
    "}\n",
    "gsearch3 = GridSearchCV(estimator = XGBClassifier( learning_rate =0.1, n_estimators=1484, max_depth=4,\n",
    " min_child_weight=12, gamma=0, subsample=0.8, colsample_bytree=0.8,\n",
    " objective= 'binary:logistic', nthread=4, scale_pos_weight=1,seed=27), \n",
    " param_grid = param_test3, scoring='roc_auc',n_jobs=4, cv=5)\n",
    "gsearch3.fit(train[predictors],train[target])\n",
    "gsearch3.cv_results_, gsearch3.best_params_, gsearch3.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "f85f6cb7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "      train-auc-mean  train-auc-std  test-auc-mean  test-auc-std\n",
      "0           0.971057       0.000238       0.971007      0.001088\n",
      "1           0.980660       0.000952       0.980480      0.001087\n",
      "2           0.981464       0.000747       0.981225      0.000674\n",
      "3           0.983026       0.000696       0.982838      0.001478\n",
      "4           0.983694       0.000383       0.983567      0.000928\n",
      "...              ...            ...            ...           ...\n",
      "1479        0.997219       0.000108       0.993742      0.000335\n",
      "1480        0.997222       0.000107       0.993742      0.000335\n",
      "1481        0.997223       0.000107       0.993743      0.000334\n",
      "1482        0.997224       0.000107       0.993743      0.000332\n",
      "1483        0.997225       0.000107       0.993741      0.000333\n",
      "\n",
      "[1484 rows x 4 columns]\n",
      "1484\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\admin\\anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1 0 1 ... 1 1 1]\n",
      "[0.5619081  0.02405899 0.73612547 ... 0.99961996 0.9999938  0.9999938 ]\n",
      "\n",
      "Model Report\n",
      "Accuracy : 0.9739\n",
      "AUC Score (Train): 0.997224\n",
      "AUC Score (Test): 0.982499\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "xgb2 = XGBClassifier(\n",
    "        learning_rate =0.1,\n",
    "        n_estimators=1484,\n",
    "        max_depth=4,\n",
    "        min_child_weight=12,\n",
    "        gamma=0.4,\n",
    "        subsample=0.8,\n",
    "        colsample_bytree=0.8,\n",
    "        objective= 'binary:logistic',\n",
    "        nthread=4,\n",
    "        scale_pos_weight=1,\n",
    "        seed=27)\n",
    "modelfit(xgb2, train, test, predictors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "62735823",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\admin\\anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[04:10:47] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "({'mean_fit_time': array([ 748.54200873,  748.33966718,  754.1630981 ,  758.40649624,\n",
       "          823.628093  ,  830.53890414,  836.6984724 ,  845.46243834,\n",
       "          907.3901073 ,  914.08244901,  922.15662007,  931.83708663,\n",
       "          989.34033957,  999.34366064, 1010.53597884, 1014.76679311]),\n",
       "  'std_fit_time': array([ 1.4983344 ,  5.26715419,  7.09255571,  6.78998847,  2.66227534,\n",
       "          5.41573563,  8.15318905,  5.12985675,  3.8220995 ,  7.64225501,\n",
       "          6.78529294,  5.54525504,  5.22460208, 10.31855939, 11.81541678,\n",
       "          5.15023143]),\n",
       "  'mean_score_time': array([0.93164639, 0.88159971, 0.91281428, 1.06087508, 0.98185759,\n",
       "         0.96294198, 0.90945134, 1.07278452, 0.92223482, 0.88476081,\n",
       "         0.93682547, 0.95000062, 0.97560196, 1.05783224, 0.95388198,\n",
       "         0.79203324]),\n",
       "  'std_score_time': array([0.10161406, 0.09360645, 0.13115296, 0.14353376, 0.12975513,\n",
       "         0.13066709, 0.07109229, 0.09577466, 0.05506742, 0.04910119,\n",
       "         0.05081225, 0.04250329, 0.01816334, 0.10233687, 0.05231298,\n",
       "         0.13575594]),\n",
       "  'param_colsample_bytree': masked_array(data=[0.6, 0.6, 0.6, 0.6, 0.7, 0.7, 0.7, 0.7, 0.8, 0.8, 0.8,\n",
       "                     0.8, 0.9, 0.9, 0.9, 0.9],\n",
       "               mask=[False, False, False, False, False, False, False, False,\n",
       "                     False, False, False, False, False, False, False, False],\n",
       "         fill_value='?',\n",
       "              dtype=object),\n",
       "  'param_subsample': masked_array(data=[0.6, 0.7, 0.8, 0.9, 0.6, 0.7, 0.8, 0.9, 0.6, 0.7, 0.8,\n",
       "                     0.9, 0.6, 0.7, 0.8, 0.9],\n",
       "               mask=[False, False, False, False, False, False, False, False,\n",
       "                     False, False, False, False, False, False, False, False],\n",
       "         fill_value='?',\n",
       "              dtype=object),\n",
       "  'params': [{'colsample_bytree': 0.6, 'subsample': 0.6},\n",
       "   {'colsample_bytree': 0.6, 'subsample': 0.7},\n",
       "   {'colsample_bytree': 0.6, 'subsample': 0.8},\n",
       "   {'colsample_bytree': 0.6, 'subsample': 0.9},\n",
       "   {'colsample_bytree': 0.7, 'subsample': 0.6},\n",
       "   {'colsample_bytree': 0.7, 'subsample': 0.7},\n",
       "   {'colsample_bytree': 0.7, 'subsample': 0.8},\n",
       "   {'colsample_bytree': 0.7, 'subsample': 0.9},\n",
       "   {'colsample_bytree': 0.8, 'subsample': 0.6},\n",
       "   {'colsample_bytree': 0.8, 'subsample': 0.7},\n",
       "   {'colsample_bytree': 0.8, 'subsample': 0.8},\n",
       "   {'colsample_bytree': 0.8, 'subsample': 0.9},\n",
       "   {'colsample_bytree': 0.9, 'subsample': 0.6},\n",
       "   {'colsample_bytree': 0.9, 'subsample': 0.7},\n",
       "   {'colsample_bytree': 0.9, 'subsample': 0.8},\n",
       "   {'colsample_bytree': 0.9, 'subsample': 0.9}],\n",
       "  'split0_test_score': array([0.99554479, 0.99540444, 0.9953134 , 0.99540537, 0.99556765,\n",
       "         0.99536201, 0.99533854, 0.99520396, 0.99546546, 0.99530061,\n",
       "         0.99532459, 0.99523391, 0.99571331, 0.99530608, 0.99532394,\n",
       "         0.99525396]),\n",
       "  'split1_test_score': array([0.99859201, 0.99862751, 0.99850963, 0.99853401, 0.9985388 ,\n",
       "         0.99853828, 0.99858273, 0.99855367, 0.99854511, 0.9986024 ,\n",
       "         0.99858764, 0.99858337, 0.99847412, 0.99857433, 0.99854593,\n",
       "         0.99851464]),\n",
       "  'split2_test_score': array([0.99923542, 0.99921492, 0.99922552, 0.99924005, 0.99921583,\n",
       "         0.99920275, 0.99920637, 0.99923386, 0.99925774, 0.99922761,\n",
       "         0.99919391, 0.99922667, 0.99921915, 0.9992311 , 0.99920526,\n",
       "         0.99922871]),\n",
       "  'split3_test_score': array([0.99214907, 0.99210941, 0.99224552, 0.99227064, 0.99197351,\n",
       "         0.99207958, 0.9922472 , 0.99230205, 0.99197197, 0.99211194,\n",
       "         0.99230257, 0.99221693, 0.99189742, 0.99205655, 0.99214279,\n",
       "         0.99216768]),\n",
       "  'split4_test_score': array([0.98070632, 0.98086375, 0.98081285, 0.98021294, 0.98034917,\n",
       "         0.9797488 , 0.98026854, 0.97973446, 0.97931407, 0.97977441,\n",
       "         0.97962219, 0.980362  , 0.97994388, 0.97978843, 0.97966464,\n",
       "         0.97936619]),\n",
       "  'mean_test_score': array([0.99324552, 0.99324401, 0.99322138, 0.9931326 , 0.99312899,\n",
       "         0.99298629, 0.99312868, 0.9930056 , 0.99291087, 0.99300339,\n",
       "         0.99300618, 0.99312458, 0.99304958, 0.9929913 , 0.99297651,\n",
       "         0.99290624]),\n",
       "  'std_test_score': array([0.00675551, 0.00669118, 0.00668305, 0.00691813, 0.00688363,\n",
       "         0.00708754, 0.00689534, 0.00708552, 0.0072706 , 0.00708756,\n",
       "         0.00713361, 0.00685843, 0.00703806, 0.0070803 , 0.00711656,\n",
       "         0.00722244]),\n",
       "  'rank_test_score': array([ 1,  2,  3,  4,  5, 13,  6, 10, 15, 11,  9,  7,  8, 12, 14, 16])},\n",
       " {'colsample_bytree': 0.6, 'subsample': 0.6},\n",
       " 0.9932455214231564)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# subsample and colsample_bytree\n",
    "param_test4 = {\n",
    " 'subsample':[i/10.0 for i in range(6,10)],\n",
    " 'colsample_bytree':[i/10.0 for i in range(6,10)]\n",
    "}\n",
    "gsearch4 = GridSearchCV(estimator = XGBClassifier( learning_rate =0.1, n_estimators=1484, max_depth=4,\n",
    " min_child_weight=12, gamma=0.4, subsample=0.8, colsample_bytree=0.8,\n",
    " objective= 'binary:logistic', nthread=4, scale_pos_weight=1,seed=27), \n",
    " param_grid = param_test4, scoring='roc_auc',n_jobs=4, cv=5)\n",
    "gsearch4.fit(train[predictors],train[target])\n",
    "gsearch4.cv_results_, gsearch4.best_params_, gsearch4.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "8c242486",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\admin\\anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[14:06:37] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=5,\n",
       "             estimator=XGBClassifier(base_score=None, booster=None,\n",
       "                                     colsample_bylevel=None,\n",
       "                                     colsample_bynode=None,\n",
       "                                     colsample_bytree=0.6,\n",
       "                                     enable_categorical=False, gamma=0.4,\n",
       "                                     gpu_id=None, importance_type=None,\n",
       "                                     interaction_constraints=None,\n",
       "                                     learning_rate=0.1, max_delta_step=None,\n",
       "                                     max_depth=4, min_child_weight=12,\n",
       "                                     missing=nan, monotone_constraints=None,\n",
       "                                     n_estimators=1484, n_jobs=None, nthread=4,\n",
       "                                     num_parallel_tree=None, predictor=None,\n",
       "                                     random_state=None, reg_alpha=None,\n",
       "                                     reg_lambda=None, scale_pos_weight=1,\n",
       "                                     seed=27, subsample=0.6, tree_method=None,\n",
       "                                     validate_parameters=None, verbosity=None),\n",
       "             n_jobs=4,\n",
       "             param_grid={'colsample_bytree': [0.75, 0.8, 0.85],\n",
       "                         'subsample': [0.75, 0.8, 0.85]},\n",
       "             scoring='roc_auc')"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# subsample and colsample_bytree\n",
    "param_test5 = {\n",
    " 'subsample':[i/100.0 for i in range(75,90,5)],\n",
    " 'colsample_bytree':[i/100.0 for i in range(75,90,5)]\n",
    "}\n",
    "gsearch5 = GridSearchCV(estimator = XGBClassifier( learning_rate =0.1, n_estimators=1484, max_depth=4,\n",
    " min_child_weight=12, gamma=0.4, subsample=0.6, colsample_bytree=0.6,\n",
    " objective= 'binary:logistic', nthread=4, scale_pos_weight=1,seed=27), \n",
    " param_grid = param_test5, scoring='roc_auc',n_jobs=4,cv=5)\n",
    "gsearch5.fit(train[predictors],train[target])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "5efd7985",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\admin\\anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1224: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[15:42:43] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.5.1/src/learner.cc:1115: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'binary:logistic' was changed from 'error' to 'logloss'. Explicitly set eval_metric if you'd like to restore the old behavior.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "({'mean_fit_time': array([812.81040015, 862.51195869, 811.7281651 , 780.75140615,\n",
       "         645.04114671]),\n",
       "  'std_fit_time': array([ 23.60141625,   5.64139992,  34.53603714,   2.12911998,\n",
       "         204.03528189]),\n",
       "  'mean_score_time': array([3.92816186, 3.65320826, 2.31476493, 1.01970897, 0.85922222]),\n",
       "  'std_score_time': array([2.04482127, 2.32196406, 2.04028899, 0.07107676, 0.19822803]),\n",
       "  'param_reg_alpha': masked_array(data=[0, 0.001, 0.005, 0.01, 0.05],\n",
       "               mask=[False, False, False, False, False],\n",
       "         fill_value='?',\n",
       "              dtype=object),\n",
       "  'params': [{'reg_alpha': 0},\n",
       "   {'reg_alpha': 0.001},\n",
       "   {'reg_alpha': 0.005},\n",
       "   {'reg_alpha': 0.01},\n",
       "   {'reg_alpha': 0.05}],\n",
       "  'split0_test_score': array([0.99554479, 0.995498  , 0.99543291, 0.99561399, 0.99576716]),\n",
       "  'split1_test_score': array([0.99859201, 0.99858212, 0.99857269, 0.99858684, 0.99858972]),\n",
       "  'split2_test_score': array([0.99923542, 0.99923541, 0.99923938, 0.9992186 , 0.9991937 ]),\n",
       "  'split3_test_score': array([0.99214907, 0.99226044, 0.99217492, 0.99211482, 0.99220565]),\n",
       "  'split4_test_score': array([0.98070632, 0.98062124, 0.98022546, 0.98082633, 0.9806261 ]),\n",
       "  'mean_test_score': array([0.99324552, 0.99323944, 0.99312907, 0.99327212, 0.99327647]),\n",
       "  'std_test_score': array([0.00675551, 0.006779  , 0.00692373, 0.00671309, 0.00679156]),\n",
       "  'rank_test_score': array([3, 4, 5, 2, 1])},\n",
       " {'reg_alpha': 0.05},\n",
       " 0.9932764676762286)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# alpha (regularization paramter)\n",
    "param_test7 = {\n",
    " 'reg_alpha':[0, 0.001, 0.005, 0.01, 0.05]\n",
    "}\n",
    "gsearch7 = GridSearchCV(estimator = XGBClassifier( learning_rate =0.1, n_estimators=1484, max_depth=4,\n",
    " min_child_weight=12, gamma=0.4, subsample=0.6, colsample_bytree=0.6,\n",
    " objective= 'binary:logistic', nthread=4, scale_pos_weight=1,seed=27), \n",
    " param_grid = param_test7, scoring='roc_auc',n_jobs=4, cv=5)\n",
    "gsearch7.fit(train[predictors],train[target])\n",
    "gsearch7.cv_results_, gsearch7.best_params_, gsearch7.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f4d30385",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}