It is ok if I compare the classification rate based on the different training and test set? 

Similar questions and discussions