skore.cross_validate

skore.cross_validate(*args, project: Project | None = None, **kwargs) → dict

Evaluate estimator by cross-validation and output UI-friendly object.

This function wraps scikit-learn’s cross_validate() function, to provide more context and facilitate the analysis. As such, the arguments are the same as scikit-learn’s cross_validate function.

The dict returned by this function is a strict super-set of the one returned by scikit-learn’s cross_validate().

For a user guide and in-depth example, see Enhancing cross-validation.

Parameters:

*args

Positional arguments accepted by scikit-learn’s cross_validate(), such as estimator and X.

projectProject, optional

A project to save cross-validation data into. If None, no save is performed.

**kwargs

Additional keyword arguments accepted by scikit-learn’s cross_validate().

Returns:

cv_resultsdict

A dict of the form returned by scikit-learn’s cross_validate() function.

Examples

>>> def prepare_cv():
...     from sklearn import datasets, linear_model
...     diabetes = datasets.load_diabetes()
...     X = diabetes.data[:150]
...     y = diabetes.target[:150]
...     lasso = linear_model.Lasso()
...     return lasso, X, y

>>> project = skore.load("project.skore")  
>>> lasso, X, y = prepare_cv()  
>>> cross_validate(lasso, X, y, cv=3, project=project)  
{'fit_time': array(...), 'score_time': array(...), 'test_score': array(...)}