Wrapper to estimate model-agnostic variable importance for multi-response models.

mrVip(yhats, X)

Arguments

yhats

A list is the list generated by mrIMLpredicts

X

A dataframe is the feature data

Details

Calculates variable importance using based on partial dependencies but could do permutations as well (more memory intensive). Key input is the object created by MrIML. Can be plotted with the plot_vi function.