- Since features are highly correlated, it is very difficult to estimate the individual feature importance using permutation_importance directly on the feature data. Rather a dimensionality reduction method is being used and importance is calculated directly on the Principal Components. This still gives an idea of the model performance.
- Add a function to extract the features most important per Principal Component. function is called `get_names_of_top_n_features_from_selected_pc()`
- Add a unit test for the `get_names_of_top_n_features_from_selected_pc()` function.