python数据挖掘--评估指标

数据挖掘路上滴滴点点，记录下常用评价指标。持续更新。

Acc、f1、recall、precision

metrics_lr = {
    'accuracy': accuracy_score(prediction_lr, Y_test),
    'f1': f1_score(prediction_lr, Y_test, average="macro"),
    'recall': recall_score(prediction_lr, Y_test, average="macro"),
    'precision': precision_score(prediction_lr, Y_test, average="macro")
}
print(metrics_lr)

 

ROC、AUC用法

lg_y_score = lg.decision_function(X_test)

lg_fpr, lg_tpr, lg_threasholds = metrics.roc_curve(Y_test.ravel(),lg_y_score.ravel())