AUC for many independent variables

[MarcB]

Hi again Adrian,

Sorry to insist, I am new to KNIME ans I am having a hard time trying to understand what comes out and goes in in each node. I have made those changes (see screenshots of the Column filter and Joiner nodes configurations), but now I end up with twice the number of rows, and still many AUC <0.50 and some of them 0.00. What am I doing wrong?

Thank you,

[nemad]

I don’t see how the changes you made would cause the number of rows to double.
As for the AUC, with your sample I also get a number of predictors below 0.5, so some might actually be really bad.
Here is the updated sample workflow you sent, if you put your data in it, you should be good to go (provided it has the same format, otherwise you will have to adjust the node settings):
Sample.knwf (1.8 MB)

[MarcB]

Hi Adrian,

It run smoothly now and results make sense. Some values are still slightly below 0.50, but I guess that’s because of sampling variations in a small test set.

I hope this helped other newcomers to KNIME as much as it helped me understanding some of its particularities. Thank you very much, I highly appreciate your support.

Best regards,
Marc