Scicit-learn Custom Scoring Function

Question

I am currently using the scikit-learn development branch: 0.15-git.

Attempting to initialize an object RidgeClassifierCVusing a custom scoring function does not currently work with an error message TypeError: __init__() got an unexpected keyword argument 'scoring'.

model = RidgeClassifierCV(scoring=make_scorer(score_func))

According to docs , the RidgeClassifier takes a parameter scoring. But according to the signature of the function, the argument score_func. However, passing the output sklearn.metrics.maker_scoreras score_funcalso fails. Any ideas?

The ultimate goal is to get RidgeClassifierCV working with multiclassic (one against all) roc auc points used for the scoring function.

Answer 1

score_func scikit-learn, y_true, y_pred estimator, X, y_true. , , , Ridge, .

score_func , . 0.15 . , , , .