SVMWeighting

asiulana

Hi everyone!

My question is about how SVMWeighting works exactly.

I went to see the java code but i didn't quite understand it all.

Does anyone know how it really works, could anyone write here a brief description but more descriptive than the rapidminer manual??

Or maybe the pseudocode for this algorithm?

Thanks in advance!!

Ana Luísa

land

Hi Ana Luísa,
SVMWeighting is very simple. As the description states, it uses the coefficients of a linear SVM hyperplane as weights for the attributes. If more than one class is given, a one versus all approch is used to gain all coefficients which are then averaged.

The more complex aspect of this algorithm hides behind the linear SVM...

Greetings,
  Sebastian