It looks like you're new here. Sign in or register to get started.
Here's a screen shot of the results tab:
are you sure that all performances going in are not NAN or Inf?
1- Input performances are precision and recall which are shown correctly when connected separately to the output.
2- The exact same setup works well for 10-fold cross-validation, but this happens when doing LOO or let's say 50-fold CV. (dataset size is around 300-400)