Logical regression comparison with R

TGT

Hi for everyone.

A Little bit confused. I perform Logical regression, in RapidMiner. it’s not clear how to choose a reference and how to look at it, but I sort of figured it out with "remap binominals". Never mind. That part my output in RapidMiner:

It looks pretty reliable, Point (label) my question is Surgery Yes or No, so for example when Age rise, ratio to Surgery- yes descend - if I understand well my output.
Now same set, same logistic regression in R version 3.5.3 (2019-03-11):
part of my output in R: 

_________________________
looks too much different! Look at Age for example . Whats wrong?

Telcontar120

I have previously noticed other problems with the logistic regression operator, as noted in the following thread: https://community.rapidminer.com/discussion/53084/logistic-regression-operator-and-weights#latest
I am not sure why that thread was closed without any apparent resolution, or whether that issue is related to the issue that you have highlighted here.
@mschmitz have you looked into the performance of the logistic regression operator and noticed any of the issues described in these threads?

MartinLiebig

Hi,

Please note that we updated the H2O backend for 9.7. This may change results.

~Martin