I have a set of dependent variables that are numeric / continuous non linear data. My independent variable (response variable) is also a continuous data but I have binarized it and made it ordinal data. This is a typical classification problem but due to some reason I want to use SVR (regression - not SVC) to predict my response variable. Is it principally correct to use SVR for a non-continuous response variable ?
Generally speaking, it is possible. In my opinion, accuracy (recall, precision, etc.) for SVC using will be better. A few years ago I have solved opposite task - to use SVC for continuous response variable (article attached):
Conference Paper Critical Zone Recognition: Classification VS. Regression (PHM - 2014)
Sergey Porotsky - Thank you so much for this insight. I was desperately looking for a logical answer and your paper provides just what I was looking for.
God bless you and all the best !
- Badges
- Science topic
- Similar topics
- Mathematics
- Applied Mathematics