Catboost: Categorical variables dominating feature importance

[Rajat700]

Problem: {I am solving a regression problem using Catboost regressor. Everytime categorical features are dominating the feature importance. Numerical variables are getting shadowed. If I discretize a numerical variable into some categories then that feature also starts coming at the top of feature importance. Can anyone explain why this might be happening. Is there any parameter that I can tune so that other numerical features don't get shadowed? }
catboost version: {0.9.1.1}
Operating System: {Windows 10}
CPU: {CPU}

[dimileeh]

Try setting the L2 parameter in the range of 30-50 and see if helps to decrease your categorical features importance and improve the score on your validation set.

[Rajat700]

I tried L2 parameter but it did not have any affect. But when I used one_hot_max_size parameter then my numerical features importance increased and score on validation set also improved.

[Sandy4321]

dlihhats , can you answer on this ?