grid_search

%%time
from sklearn.model_selection import GridSearchCV

model_gs = RandomForestClassifier(random_state=12345, criterion="gini")
parametrs_gs  = {
    'n_estimators': range(1,126, 25),
    'min_samples_leaf': range (1,8),
    'max_depth': range (1,13, 2)
}
gs = GridSearchCV(estimator = model_gs,
                  param_grid = parametrs_gs,
                  cv=5,
                  scoring=['f1', 'roc_auc'],
                  refit='f1')
gs.fit(pd.concat([features_train, features_valid]), pd.concat([target_train, target_valid]))

print(gs.best_estimator_)
print(gs.best_score_)

    #Output
    #RandomForestClassifier(max_depth=11, min_samples_leaf=2, n_estimators=76,
                       #random_state=12345)
    #0.866125
    #CPU times: user 3min 23s, sys: 932 ms, total: 3min 24s
    #Wall time: 3min 25s

# ПРИМЕНЯЮ ПАРАМЕТРЫ НА МОДЕЛИ
%%time
model = RandomForestClassifier(random_state=12345, criterion='gini', max_depth=11, min_samples_leaf=2, n_estimators=76)
model.fit(features_train, target_train)
predicted_train = model.predict(features_train)
predicted_valid = model.predict(features_valid)
# print(f1_score(target_train, predicted_train))
print("F1: ", f1_score(target_valid, predicted_valid))
print("ROC_AUC: ", roc_auc_score(target_valid, predicted_valid))
print("RECALL: ", recall_score(target_valid, predicted_valid))
print("PRECISION: ", precision_score(target_valid, predicted_valid))
print("Accuracy_score ", accuracy_score(target_valid, predicted_valid))

#Output
#F1:  0.5466448445171849