Tag: shap

  • Shap or LGBM gain for feature selection?

    Shap or LGBM gain for feature selection? Which one do you use during recursive feature elimination or forward/backward selection? I’ve always used gain and only used shap for analytics on model predictions, but came across some shap values recommendations. Bonus question: have you used “null importance” / permutation method? Fitting models with shuffled targets to…