Re: customized cross validation in weka?

>         Is there a user-defined or cutomised cross validation method
> in Weka? I mean,  for example, I have 1000 data records and I want to
> do a 5-fold CV, I want to take recordes from 1 to 200 as fold1,
> 201-400 as fold2 , ..., 801-1000 as fold5. Can I let Weka do cross
> validation on these self-defined folds?

If you're trying to avoid the randomization/stratification of the data,
there's no way around. You will have to implement that yourself. Check out
the  following method of the weka.classifiers.Evaluation class:
  crossValidateModel(Classifier,Instances,int,Random)

Commenting out the randomize and stratify method calls should give you
your desired custom cross-validation. Whether these results are then
reliable, that's a different question...

NB these changes don't affect the Explorer and the Experimenter, only when
you call a classifier from the commandline!

HTH

Cheers, Peter
--

-- 
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/     +64 (7) 838-4466 Ext. 5174