While running the SMO classifier in weka, if I have inputted my training labels as 0 and 5, (A binary set), then while running the classifier model on test data, are the outputs some decimal values between 0 and 5 which get distinctly classified into the two binary labels at a latter step, or are there no intermediate decimal values?

If they exist, how to obtain these intermediate decimal values?

(Eg, In the above model, does the smo/svm classifier assign values like 1 , 2, 3 and 4, or some other decimal/float value within the given range and then these get appropriately grouped under the 0 and 5 value classes).

  • 1
    $\begingroup$ It depends on weka, what kind of assumptions are made with your data-input, but it i'm pretty sure, that it just recognizes, that your data consists of two classes. Then your labels don't matter. The output values just describe the distance to the next support-vector (near class A or near class B). The idea you mention sounds more like a regression model, which is completely different from classification! If you are using more than two classes, there are direct multi-class formulations (crammer-singer), but usually a one-vs-rest / one-vs-all approach is used->multiple distances combined $\endgroup$ – sascha Jun 26 '16 at 22:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.