How to do feature selection based on permutation importance in Pyspark?

Question

I\'m building a model with the mmlspark lightGMB model. I\'m trying to calculate permutation feature importance for it (in order to do feature selection).

I understand th