sklearn make_scorer needs_proba=True时输入形状不好

Question

下面的代码可以正常运行:
scorer = make_scorer(roc_auc_score)

但给出"ValueError: bad input shape":
scorer = make_scorer(roc_auc_score, needs_proba = True)

代码是:
clf = GaussianNB() cv = ShuffleSplit(features.shape[0], n_iter = 10, test_size = 0.2, random_state = 0) scorer = make_scorer(roc_auc_score, needs_proba = True) score = cross_val_score(clf, features, labels, cv=cv, scoring=scorer)

我如何解决这个错误，使分数是基于概率估计?

Answer 1

如果你正在使用一个默认的评分指标，你不需要传递一个可调用的cross_val_score，你可以，你可以只是调用它与你正在使用的指标的名称:

score = cross_val_score(clf, features, labels, cv=cv, scoring='roc_auc_score')