I think I found a minor bug in the XGBoost Tree Ensemble Learner Node. My ID column has thousands of distinct values and should not be used as a predictor in that node, hence I excluded it in the column selection. The learner works perfectly, but when I re-configure that learner after that, it shows that my ID column is among the selected predictors. And when I deselect that column and hit “OK”, KNIME tells me that nothing has changed and the node will therefore not be reset.
I agree that sounds like a bug. Unfortunately, I can’t reproduce it in my installation, so I will need a bit of help in figuring it out.
Are there any errors displayed in the KNIME log or the console?
Thank you for playing with the new nodes
Hello @nemad, I had a look at the log and console but I didn’t gain any insights. Maybe you can try this workflow to reproduce the bug. (I use Windows 10.) BUG Xgboost tree ensemble.knwf (12.3 KB)
Thank you for your quick reply, your workflow was very helpful as I can now reproduce the problem.