I'm trying to generate the interpretability of a large dataset with 177000 features. Specifically I am doing permutation using the permutation_importance method from scikit-learn.

I'm using a machine with 16GB of ram and 4 cores and it's taking a lot of time more than two days.

I call the method with the following line of code:

permutation_importance(my_model, xtr, ytr, scoring=roc_auc, random_state=2020)

Does anyone know of a more effective way to speed up the swapping process with this library? And how to report the progress of the calculations?