In liblinear and libsvm, the option -v k allows us to run k-fold cross validation. But to test statistical significance, I need the accuracy obtained on each fold. Of course there is the long drawn way to create each fold and then run train and test k times, but I was wondering if there is a way to get the accuracy of each fold even when I use the -v option.
I think liblinear and libsvm will have the same answer to my question. Please correct me if I am wrong about this.
