According to this practical guide, it seems the number of folds should be selected according to a "rule of thumb". This means for binary traits in the UKBB some cross validations should use more folds.
Implementation idea:
Add default selection based on n-eff that is overided by the estimator file specification.
According to this practical guide, it seems the number of folds should be selected according to a "rule of thumb". This means for binary traits in the UKBB some cross validations should use more folds.
Implementation idea: