2
$\begingroup$

Suppose I am about to use SVM for learning a classification or ranking function. Suppose that my feature vectors are two dimensional and that values for one dimension are, say, natural numbers and the values for the other dimension are real numbers in $[0,1]$. Can this difference in magnitudes somehow negatively affect learning? Are there some guidelines for scaling feature values? Any references appreciated.

$\endgroup$
1
$\begingroup$

I don't see how there should be any negative consequences of this. To see why, you can "vectorize" your feature matrix so that you still have scalars for each "new" feature. That is, instead of a matrix of n features each with k values, you have a vector of length n×k with scalar features. SVMs with a mixture of categorical/integral + real components is standard.

$\endgroup$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.