Pages: [1]
  Print  
Author Topic: [SOLVED] Select Hyperparameters of SVM in cross-validation?  (Read 900 times)
johnny5550822
Newbie
*
Posts: 7


« on: March 04, 2014, 08:38:51 PM »

Hi,

I want to clarify one thing about what rapidminer exactly is doing. When I put a SVM module inside cross-validation (e.g. 10-fold), will the SVM algoirithm optimize the hyperparameters (e.g. C) based on cross-validation result?

If so, what about if I don't have validation, and basically just give data to SVM module, how does rapidminer get the hyperparameters value?

Thanks,
Johnny

« Last Edit: March 11, 2014, 02:01:43 PM by Marius » Logged
fras
Global Moderator
Jr. Member
*****
Posts: 86


« Reply #1 on: March 05, 2014, 01:11:13 PM »

Cross Validation (CV) does not optimize at all. Doing a 10-fold or 5-fold CV ensures only to get performance parameters you can trust.
If you want to optimize e.g. "C" you have to put the CV into the "Optimize Parameters" Operator that performs a training/validating with
all selected values of C. Take a look into the example process delivered together with the operators help.
Logged
johnny5550822
Newbie
*
Posts: 7


« Reply #2 on: March 11, 2014, 02:13:04 AM »

Got it, thanks!
Logged
Pages: [1]
  Print  
 
Jump to: