Hi all,
I was trying to do a sentiment prediction using KNMIE text processing nodes. Beside getting pretty bad results on the testing set (maybe I missed a step?) I saw some weird columns in the tf matrix
Can anyone elaborate on this?
Also I sometimes notice some issues / warnings in the console that something could not converge. Is there a documentation (beside a finished workflow on the hub) for that?
Thanks and best
The column names seem like an UTF/ISO encoding problem, and the convergence message is probably a result of of something going sideways with the Logistic Regression Learner.
Apart from that, it’s hard to say why you are not getting the results you expect based on the screenshot. If you can post your workflow along with a small example dataset, that would help with the troubleshooting process.