CODEWITHSUNDEEP
machine-learningwekacross-validation
gopalkoduri
gopalkoduri

Reputation: 442

Does weka balance the train/test sets across classes in cross validation?

When I perform a n-fold cross validation on m classes, in each fold, are the train and test sets balanced? By balanced, I mean to ask if there are (nearly) the same set of samples from each class in train and test sets.

Upvotes: 0

Views: 2121

Answers (1)

Steve Tjoa
Steve Tjoa

Reputation: 61114

If I understand this part correctly:

I mean to ask if there are (nearly) the same set of samples from each class in train and test sets.

... I think you are referring to stratified cross validation: where the class proportions for each fold is the same as the class proportions of the entire data set.

Yes, it is certainly possible in Weka, but you may want to consult the documentation. Might these help?

http://weka.wikispaces.com/Primer

http://weka.wikispaces.com/Generating+cross-validation+folds+(Java+approach)

http://weka.sourceforge.net/doc/weka/filters/supervised/instance/StratifiedRemoveFolds.html

Upvotes: 5

Related Questions