[R] to determine the variable importance in svm

[mxkuhn]

> is ipred:errorest method is good enough for validating (or cross check) my
> svm result?

Yes, if you know what values of the tuning parameters to use, but I don't know why it was failing.

Max