search

ecpolley / SuperLearner

Current version of the SuperLearner R package
272 stars 72 forks source link

Name consistency in SL.polymars #146

Open bdwilliamson opened 1 year ago

bdwilliamson commented 1 year ago

For continuous outcomes, SL.polymars returns list(pred = pred, fit = list(object = fit.mars)), whereas for binary outcomes, SL.polymars returns list(pred = pred, fit = list(fit = fit.mars)). I think the former is used throughout the rest of SuperLearner, so it would be great to have the binary-outcome polymars behave similarly. Thanks!

ecpolley commented 1 year ago

I believe this traces back to the different polymars predict methods, polspline::predict.polymars uses object, while polymars::ppolyclass uses fit. Let me check if there would be any adverse consequences of the change to align with object.