# XGBClassifier
print('XGBClassifier--------------------------------------')
param_grid = {
    'max_depth':[10,12],
    'min_child_weight':[3,5],
    'gamma':[1,2],
    'subsample':[0.75,0.8],
    'colsample_bytree':[0.75,0.8]
}
# clf = XGBClassifier(learning_rate =0.01)
grid_search = GridSearchCV(clf, param_grid, n_jobs=-1, cv = skf, verbose=2, scoring='roc_auc').fit(projected, ytrain)
print('The best parameters are :', grid_search.best_params_)
print(grid_search.best_score_)
clf = XGBClassifier(learning_rate =0.01,max_depth=grid_search.best_params_['max_depth'],min_child_weight=grid_search.best_params_['min_child_weight'],gamma=grid_search.best_params_['gamma'],subsample=grid_search.best_params_['subsample'],colsample_bytree=grid_search.best_params_['colsample_bytree'])
accuracies, auc_rocs, auc_prs = train_K_Fold(projected, ytrain, clf, skf)
accuracy, auc_roc, auc_pr = get_means_std(accuracies, auc_rocs, auc_prs)