I would like to compare the "standard" rFerns' model output with one model's output from a modified version of rferns (only living in my laptop).
The two rFerns were fitted to a multilabel dataset (birds from mldr.datasets package) and because of that I don't have the "importance" feature. What any option do I have?
Many thanks in advance!!
Sorry for a late reply.
Unfortunately, importance in multi-label case is still a topic for further research; for now, you can collect importance from a battery of one-class-vs-other models and combine them somehow.