Bug in train panel when parameter of learner is changed and trained again
Closed this issue · 2 comments
florianfendt commented
Bug in train panel when parameter of learner is changed and trained again
Coorsaa commented
I'm still facing the following bug:
when I train a model without probailitiy estimation and predict it. I get a data table with only truth and response as it should be. If I decide afterwards, to set prob to "yes", train the model again, and predict it afterwards, I still get only truth and response values, but no probabilities.
florianfendt commented
This bug is super weird to me. Did you notice it only happens for the randomForest? The other learners are all fine (I checked like 5-6)