My data calculated from Matlab has a lot of NaN for some correlation functions. My classification prediction results are terrible, so I was wondering if this was due to a problem. Will LIBSVM accept NaN as a valid input?
No one has answered this question yet.
See related questions: