hack_dev.py 文件源码

def hackathon_GBC_model(clf, train, features):
    clf.fit(train[features], train["Class"])
    probab_of_predict = clf.predict_proba(train[features])[:,1]
    predict_train = clf.predict(train[features])
    cv_score = cross_val_score(clf, train[features], train["Class"], cv=5, scoring="roc_auc")
    print("----------------------Model performance-----------------------")
    print("Accuracy score: ", accuracy_score(train["Class"].values, predict_train))
    print("AUC: ", roc_auc_score(train["Class"],probab_of_predict) )
    print("CV score: Mean - {}, Max - {}, Min - {}, Std - {}".format(np.mean(cv_score), np.max(cv_score),
                                                                     np.min(cv_score), np.std(cv_score)))

    Relative_Feature_importance = pd.Series(clf.feature_importances_, features).sort_values(ascending=False)
    Relative_Feature_importance.plot(kind='bar', title='Order of Feature Importance')
    plt.ylabel('Feature Importance')
    plt.show()