Unfortunately, a bug crept in regarding the Experimenter in combination with cross-validation. Instead of summing up the differents folds of a run, they're listed as separate "classifiers" (prefixed with a number).
There a two ways to fix this:
Drawback: you have to do that whenever you select a new experiment.
This fixes the wrong indices in the default Experimenter.props file in the weka.jar archive.
Sorry about that, new versions of Weka will be released in the near future!
Log in to post a comment.