Dear R-society,
I know this is probably a question to LibSVM, but I have not found a
LibSVM mailing list, so I would be really grateful if someone from this
list helped me.
In the LibSVM documentation I have found a passage about dealing with
unbalanced sets, that says that in this case one sets two different
regularization parameters C^{-} and C^{+}. However I am not sure about how
this corresponds to the parameters in the svm function. There I find C and
class.weights. So am I right supposing that
C^{-} = class.weights[1]*C and C^{+} = class.weights[2]*C ?
--
Sincerely yours,
Yulia Matveyeva,
Department of Statistical Modelling,
Faculty of Mathematics and Mechanics,
St Petersburg State University, Russia
[[alternative HTML version deleted]]