Closed gverbock closed 2 years ago
@Matgrb @timvink I made a PR on what I believe is a bug in the implementation of the mask. It introduces fluctuations in the shap value between for example train and test.
I believe one could also go further and make the masking optional for algorithm that are not Linear Regressions.
Yes, this is finished. I will remove the draft status.
@Matgrb Can I merge, I see you approved so I guess it is ready to go.
Pull request based on #176