Out of sample metrics

[AhmedThahir]

Hi there!
Great work on this project.

Is there a way to use custom out-of-sample metrics such as accuracy to inspect mean decrease in accuracy with cross validation, rather than only using feature importance which is in-sample?

• If yes, kindly share the documentation, as I wasn't able to find it.
• If no, kindly share a starting point on how I can extend the current Boruta implementation.

On a sidenote, I request you to open up the discussions tab so that such questions do not come up as issues.

[AhmedThahir]

Any updates?