Hi, I have an experimental (fMRI) paradigm wherein the conditions were balanced across runs, and performed a searchlight MVPA analysis with leave-one-run-out cross validation. In some subjects, however, we found excessive head motion and had to eliminate scans, which has resulted in some condition imbalance across runs.
There are two options that occur to me to resolve this, and in each case, I would use mvpa2.generators.resampling.Balancer. 1) Attempt to reconstruct validation folds of the original size, possibly reducing the number of folds. 2) Attempt to reconstruct the same number of folds, but with fewer instances of each condition. I don't have a principled way to choose between these two options, and it's possible that I've missed other, better options. I'd appreciate any insight you might have on the advantages of each strategy and any considerations that might come into play. Thanks, -- Chris Johnson
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Pkg-ExpPsy-PyMVPA mailing list [email protected] http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-exppsy-pymvpa
