Does CatBoost support L1 regularization of the cost function?

Question

Does CatBoost suppport L1 regularization? The hyper parameter l2_leaf_reg controls the L2 regularization term of the cost function, but is there any way to use the L1 norm?

Answer 1

if you look at the documentation it doesn't seem like it Catboost reference params

I guess if you are interested in L1 to perform feature selection, a possible solution would be to run Lightgbm or XGBoost first with L1 regularization to get the relevant subset of features and then pass those to Catboost