-1

I am using WEKA on my project. I use SVM(SMO) to classify the result. I firstly collected training set data and ran with 10fold crossvalidation. Then I built a classifier with those training set and made it ran in real-time, means real-time extracting features and passing them into the classifier. However, the classifier returns 0 at most cases( we have 5 labels, 0 is the first label)? It may return other label, but very rare.

What might be the possible reason that cause the classier keep outputting 0? Thanks a lot.

Foreverniu
  • 349
  • 1
  • 3
  • 18

1 Answers1

1

The most probable reasons are:

  • You trained such a trivial model, probably because you did not provide correct parameters for SVM for your particular dataset (which cannot be just "guessed", you need to find them)
  • You have some kind of implementation error

nothing more can be told given such brief description of the problem.

lejlot
  • 64,777
  • 8
  • 131
  • 164