Samuel Wilson

I initially thought, since multiclass uses the `softmax`, that the initial scores might be the result of using `softmax` on the `logit(% of total classes)` for each class. But that...

Interesting, from my understanding the multiclass procedure creates N binary trees and then takes the softmax of their output. Therefore, this behavior seems strange. For larger class sizes, this will...

@KaikeWesleyReis any update on this? This is a severe issue if you can reproduce.

Hmmm are the imputed values for that second variable (top right in the chart) some constant value? I could see this error coming from trying to get the kernel smoothing...

Thanks, well that's definitely a bug - I'll have to make it skip any imputations with 0 variance.

Link to scipy issue. Temporary fix is to use `balanced_tree=False`. https://github.com/scipy/scipy/issues/14799

Feel free to send a pull request with the changes, but I don't think this is something I would implement myself.

Does this pass a local R CMD check? It looks like travis CI fails because xgboost isn't available... What a pain.

Okay I'll pull this down and run the CMD checks - if it's just Travis that's the problem we should be good to go.

@kapsner Sorry for the late reply, it's crunch time at work right now. Would you like to be a maintainer of this project? I'm afraid I can't give this project...