How to combine Parameter Optimization and Cross Validation

KNIME Analytics Platform
#1

Hi,

I’d like to use a parameter optimization loop on top of cross validation but I can’t seem to get it to work in a workflow. Basically, I’m trying to run a series of parameter combinations each through a 10-fold cross validation and return the optimal set of parameters. Any suggestions? Not sure if this is already covered in an example…

Thanks,

#2

Hi macsmith,

Please find attached an example of training an SVM model using a 5-fold cross validation combined with parameter optimisation.

Best,
Anna
Cross_Validation_with_SVM_and_Parameter_Optimisation.knar.knwf (53.7 KB)

2 Likes
#3

So simple. Thanks for this. I was running into trouble because my cross-validation was contained inside a metanode and I was having difficulties connecting the start and end loops through the metanode ports (and i think i had the scorer out of place).

#4

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.