11_tuning_gbm.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "d01fd7cb",
   "metadata": {},
   "source": [
    "# 11. Tuning GBMs"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2158bdab",
   "metadata": {},
   "source": [
    "Let's <span style=\"color:#173F5F\">**experiment**</span> with the classification example (```data = dfc```) to get even more grip on the tuning of <span style=\"color:#173F5F\">**GBM parameters**</span>.\n",
    "\n",
    "Which ```objective``` to specify for <span style=\"color:#173F5F\">**classification**</span>?\n",
    "* ```\"binary:logistic\"```: logistic regression for 0-1 outcomes\n",
    "* ```\"binary:logitraw\"```: logistic regression for 0-1 outcomes (before transformation, scores are computed)\n",
    "* ..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "5c6edbd0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# import relevant libraries\n",
    "from sklearn.model_selection import GridSearchCV,RandomizedSearchCV,train_test_split\n",
    "from matplotlib.colors import ListedColormap\n",
    "from matplotlib import pyplot as plt\n",
    "from matplotlib import animation\n",
    "import xgboost as xgb\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "4fcdcd46",
   "metadata": {},
   "outputs": [],
   "source": [
    "x1 = np.tile(np.linspace(start = 0.1, stop = 10, num=int(10/0.1), dtype = np.float32), 100)\n",
    "x2 = np.sort(np.tile(np.linspace(start = 0.1, stop = 10, num=int(10/0.1), dtype = np.float32), 100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "0f5cadeb",
   "metadata": {},
   "outputs": [],
   "source": [
    "y = np.minimum(1,\n",
    "               np.maximum(0,\n",
    "                          (1*(x1+2*x2<8) + 1*(3*x1+x2>30) + np.random.normal(0,0.5,x1.shape[0])).astype(int)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "00e8742c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<mpl_toolkits.mplot3d.art3d.Path3DCollection at 0x19a63c373c8>"
      ]
     },
     "execution_count": 4,
     "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": [
    "%matplotlib inline\n",
    "ax = plt.axes(projection='3d')\n",
    "\n",
    "# Data for three-dimensional scattered points\n",
    "zdata = x2\n",
    "xdata = y\n",
    "ydata = x1\n",
    "\n",
    "# Data for a three-dimensional line\n",
    "ax.scatter3D(xdata, ydata, zdata, c=zdata, cmap='Greens')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "c50157c3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# packing explanatory variables in a pd.DataFrame\n",
    "x = pd.DataFrame({'feature_1':x1, 'feature_2':x2})\n",
    "y = pd.DataFrame({'y':y})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "27ca07ce",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
       "              colsample_bynode=1, colsample_bytree=1, eval_metric='auc',\n",
       "              gamma=0, gpu_id=-1, grow_policy='lossguide',\n",
       "              importance_type='gain', interaction_constraints='',\n",
       "              learning_rate=0.300000012, max_delta_step=0, max_depth=6,\n",
       "              min_child_weight=1, missing=nan, monotone_constraints='()',\n",
       "              n_estimators=100, n_jobs=16, num_parallel_tree=1, random_state=0,\n",
       "              reg_alpha=0, reg_lambda=1, scale_pos_weight=1, subsample=1,\n",
       "              tree_method='hist', validate_parameters=1, verbosity=None)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#obtaining default parameters by calling .fit() to XGBoost model instance\n",
    "xgb_classifier1 = xgb.XGBClassifier(objective='binary:logistic',\n",
    "                          n_estimators = 100,\n",
    "                          booster='gbtree',\n",
    "                          eval_metric='auc',\n",
    "                          tree_method='hist',\n",
    "                          grow_policy='lossguide')\n",
    "xgb_classifier1.fit(x,y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "d6966e91",
   "metadata": {},
   "outputs": [],
   "source": [
    "#extracting default parameters from benchmark model\n",
    "default_params = {}\n",
    "gparams = xgb_classifier1.get_params()\n",
    "\n",
    "#default parameters have to be wrapped in lists - even single values - so GridSearchCV can take them as inputs\n",
    "for key in gparams.keys():\n",
    "    gp = gparams[key]\n",
    "    default_params[key] = [gp]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "46c79101",
   "metadata": {},
   "source": [
    "### 11.1. Grid Search"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4ebfbdae",
   "metadata": {},
   "source": [
    "We have two different ways to set up a grid-search approach on the hyperparameters. We present a manual and an automated procedure to perform grid-search on the hyperparameters."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a22628b4",
   "metadata": {},
   "source": [
    "##### 11.1.1. Manual Grid-search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "fafea2e2",
   "metadata": {},
   "outputs": [],
   "source": [
    "base_search_dict = {'max_depth':[1,3,5],\n",
    "                    'learning_rate':[0.01,0.1,1]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "6d5236ba",
   "metadata": {},
   "outputs": [],
   "source": [
    "# perform cartesian product over the elements of base_search_dict\n",
    "from itertools import product\n",
    "def my_product(inp):\n",
    "    return (dict(zip(inp.keys(), values)) for values in product(*inp.values()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ada7295f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# prepare the expanded dictionary to \"save\" the results\n",
    "cp_grid_search = {}\n",
    "for e,i in enumerate(my_product(base_search_dict)):\n",
    "    name_model = f'model_{e}'\n",
    "    cp_grid_search[name_model] = i"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "a3e984e1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# fit 9 XGBClassifiers and retain the prediction in the dictionary created above\n",
    "for i in cp_grid_search:\n",
    "    xgb_gs = xgb.XGBClassifier(objective='binary:logistic',\n",
    "                          n_estimators = 100,\n",
    "                          booster='gbtree',\n",
    "                          eval_metric='auc',\n",
    "                          tree_method='hist',\n",
    "                          grow_policy='lossguide',\n",
    "                          max_depth = cp_grid_search[i]['max_depth'],\n",
    "                          learning_rate = cp_grid_search[i]['learning_rate'])\n",
    "    xgb_gs.fit(x,y)\n",
    "    cp_grid_search[i]['predict_class'] = xgb_gs.predict(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "72dceff3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "model_0 {'max_depth': 1, 'learning_rate': 0.01, 'predict_class': array([0, 0, 0, ..., 0, 0, 0])}\n",
      "model_1 {'max_depth': 1, 'learning_rate': 0.1, 'predict_class': array([0, 0, 0, ..., 0, 0, 0])}\n",
      "model_2 {'max_depth': 1, 'learning_rate': 1, 'predict_class': array([0, 1, 1, ..., 0, 0, 0])}\n",
      "model_3 {'max_depth': 3, 'learning_rate': 0.01, 'predict_class': array([0, 0, 0, ..., 1, 1, 1])}\n",
      "model_4 {'max_depth': 3, 'learning_rate': 0.1, 'predict_class': array([1, 1, 1, ..., 0, 0, 0])}\n",
      "model_5 {'max_depth': 3, 'learning_rate': 1, 'predict_class': array([1, 1, 1, ..., 1, 0, 0])}\n",
      "model_6 {'max_depth': 5, 'learning_rate': 0.01, 'predict_class': array([0, 0, 0, ..., 0, 0, 0])}\n",
      "model_7 {'max_depth': 5, 'learning_rate': 0.1, 'predict_class': array([0, 1, 1, ..., 1, 0, 0])}\n",
      "model_8 {'max_depth': 5, 'learning_rate': 1, 'predict_class': array([1, 1, 1, ..., 1, 0, 0])}\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 18 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import ConfusionMatrixDisplay\n",
    "%matplotlib inline\n",
    "f, axes = plt.subplots(3, 3, figsize=(20, 20), sharey='row')\n",
    "axes = axes.ravel()\n",
    "\n",
    "for i, (key, classifier) in enumerate(cp_grid_search.items()):\n",
    "    print(key, classifier)\n",
    "    axes[i].set_title(key)\n",
    "    y_pred = classifier['predict_class']\n",
    "    disp = ConfusionMatrixDisplay.from_predictions(y, y_pred, display_labels=['0', '1'], ax=axes[i], xticks_rotation=0)\n",
    "\n",
    "plt.subplots_adjust(wspace=0.40, hspace=0.1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6bba6f38",
   "metadata": {},
   "source": [
    "##### 11.1.2. Automated Grid-search"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "01c2a562",
   "metadata": {},
   "source": [
    "We can exploit ```GridSearchCV``` from ```sklearn``` to automate the grid-search operation:\n",
    "* a set of hyperparameters is required to be defined in a dictionary a priori\n",
    "* all the combinations of the values defined in the dictionary are passed and evaluatied using Cross-Validation method\n",
    "* we define a scoring rule to define which is the best parameter combination"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "001423c9",
   "metadata": {},
   "source": [
    "The method ```sklearn.model_selection.GridSearchCV``` presents the following parameters:\n",
    "\n",
    "* ```estimator```: pass the model instance for which you want to check the hyperparameters\n",
    "* ```param_grid```: the dictionary object that holds the hyperparameters you want to try\n",
    "* ```scoring```: evaluation metric that you want to use, you can simply pass a valid string/object of evaluation metric\n",
    "* ```n_jobs```: number of processes you wish to run in parallel for this task if it -1 it will use all available processors. \n",
    "* ```cv```: number of cross-validation you have to try for each selected set of hyperparameters\n",
    "* ```verbose```: you can set it to 1 to get the detailed print out while you fit the data to GridSearchCV\n",
    "* ..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "7340a4f4",
   "metadata": {},
   "outputs": [],
   "source": [
    "# define dictionary for gridsearch\n",
    "search_dict = {'max_depth':[1,3,5],\n",
    "               'learning_rate':[0.01,0.1,1]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "d06629da",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 3 folds for each of 9 candidates, totalling 27 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\xgboost\\sklearn.py:888: 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",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:98: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n",
      "C:\\Users\\u0149060\\Anaconda3\\envs\\dev_env\\lib\\site-packages\\sklearn\\preprocessing\\_label.py:133: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=3,\n",
       "             estimator=XGBClassifier(base_score=0.5, booster='gbtree',\n",
       "                                     colsample_bylevel=1, colsample_bynode=1,\n",
       "                                     colsample_bytree=1, eval_metric='auc',\n",
       "                                     gamma=0, gpu_id=-1,\n",
       "                                     grow_policy='lossguide',\n",
       "                                     importance_type='gain',\n",
       "                                     interaction_constraints='',\n",
       "                                     learning_rate=0.300000012,\n",
       "                                     max_delta_step=0, max_depth=6,\n",
       "                                     min_child_weight=1, missing=nan,\n",
       "                                     monotone_constraints='()',\n",
       "                                     n_estimators=100, n_jobs=16,\n",
       "                                     num_parallel_tree=1, random_state=0,\n",
       "                                     reg_alpha=0, reg_lambda=1,\n",
       "                                     scale_pos_weight=1, subsample=1,\n",
       "                                     tree_method='hist', validate_parameters=1,\n",
       "                                     verbosity=None),\n",
       "             param_grid={'learning_rate': [0.01, 0.1, 1],\n",
       "                         'max_depth': [1, 3, 5]},\n",
       "             return_train_score=True, scoring='roc_auc', verbose=1)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# execute cross_validation\n",
    "xgb_classifier2 = GridSearchCV(estimator=xgb_classifier1,  \n",
    "                               param_grid=search_dict,\n",
    "                               scoring='roc_auc',\n",
    "                               return_train_score=True, \n",
    "                               verbose=1, \n",
    "                               cv=3)\n",
    "xgb_classifier2.fit(x, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "94a559a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# extract optimal parameters after grid-search\n",
    "optimal_params = {}\n",
    "gparams = xgb_classifier2.get_params()\n",
    "\n",
    "#default parameters have to be wrapped in lists - even single values - so GridSearchCV can take them as inputs\n",
    "for key in gparams.keys():\n",
    "    gp = gparams[key]\n",
    "    optimal_params[key] = [gp]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "12393ba2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'learning_rate': 0.1, 'max_depth': 3}"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# extract best parameters based on roc_auc scoring criteria\n",
    "xgb_classifier2.best_params_ # according to manual grid search, this is model_4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "66ac700a",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.7.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}