When use step function in R, the scoring criteria is AIC, so the penalty of dimension is part of the scoring. But when I use python, I don't know which scoring I should choose, especially when 'k_features' is not fixed.
When I use 'k_features' = (1,8), the result of sfs.k_feature_idx_ is always just (0,)
Can anyone tell me how to fix this?
