search

ing-bank / probatus

Validation (like Recursive Feature Elimination for SHAP) of (multiclass) classifiers & regressors and data used to develop them.
https://ing-bank.github.io/probatus
MIT License
132 stars 40 forks source link

Remove rounding from metrics added to reports. #190

Closed christophermadsen closed 1 year ago

christophermadsen commented 2 years ago

When reporting training and validation metrics, we round to 3 decimals, but this unnecessarily removes experiment granularity that some users might need. https://github.com/ing-bank/probatus/blob/f409c8e3442f096e91283b30b26ee1fda73d8b65/probatus/feature_elimination/feature_elimination.py#L573-L582

I propose that we refrain from using any unnecessary rounding and remove it from the lines above.