feature_importances doesn't work in iterated model

[Moelf]

[ablaom]

Thanks for reporting. Yes, it looks to me that feature_importance method is not forwarded to the wrapper IteratedModel.

[ablaom]

I think the same forwarding we have for the TunedModel wrapper works here.

[ablaom]

@Moelf As a workaround, can you try feature_importances(fitted_params(mach_bdt).machine), assuming mach_bdt is the machine bound to your IteratedModel instance.