Logistic Regression doesn't work with training stats

[yaeldekel]

Stats that are being computed at training time aren't passed to the predictor.

[Zruty0]

#1006 is related?

[montebhoover]

Yes, this looks the same as #1006.

[montebhoover]

@yaeldekel , from the comment on #1006 below, it sounds like we are both not passing training-time stats to the predictor, and there are more stats for us to calculate. Does that ring true?

"If I do [pass _stats to the predictor], then predictor.Statistics.GetCoefficientStatistics still returns null. It appears that you calculate the inverted Hessian, but then do not use it to calculate standard errors (and pass these in the constructor of `LinearModelStatistics')"

[Ivanidzo4ka]

what about #612 ?
I would assume you fix it because you working on one not-allowed-to-mention repository, and even if you bring stats, you wouldn't bring whole old version since this one doesn't have standard errors.