Skip to content

Check if shap accepts pd.DataFrame (and for which Explainers) #508

Open
@hbaniecki

Description

@hbaniecki

...and possibly remove the conversion from pandas to numpy in ShapWrapper.

Then, explainer.model_info['arrays_accepted'] would become redundant, which in fact fixes #507 by making the scikit Pipeline fail gracefully in the downstream package (with an appropriate exception message).

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions