我有一个分类数据集,有3个类标签 [0,1,2]
.
我想运行交叉验证并尝试几个估算器,但我对只有1级和2级精度的评分感兴趣 . 我不关心0级的精度,我不希望它的得分甩开CV优化 . 我也不关心任何课程的召回 . 换句话说,我想确保无论何时预测1或2,它都具有很高的置信度 .
所以问题是,如何运行 cross_val_score
并告诉其评分函数忽略0级的精度?
UPDATE :根据接受的答案,这是一个示例答案代码:
def custom_precision_score(y_true,y_pred):
precision_tuple, recall_tuple, fscore_tuple, support_tuple = metrics.precision_recall_fscore_support(y_true, y_pred)
precision_tuple = precision_tuple[1:]
support_tuple = support_tuple[1:]
weighted_precision = np.average(precision_tuple, weights=support_tuple)
return weighted_precision
custom_scorer = metrics.make_scorer(custom_precision_score)
scores = cross_validation.cross_val_score(clf, featuresArray, targetArray, cv=10, scoring=custom_scorer)
1 回答
cross_val_score
包含scorer callable object,使用make_scorer
可以set with your own test strategy . 并且您可以在自定义分数函数score_func(y, y_pred, **kwargs)
中设置要测试的组,该函数由make_scorer
调用 .