🎉 🎉 RAPIDMINER 9.10 IS OUT!!! 🎉🎉

Download the latest version helping analytics teams accelerate time-to-value for streaming and IIOT use cases.

CLICK HERE TO DOWNLOAD

Optimisation Grid - Model Parameters for Training and Testing

LobbieLobbie Member Posts: 10 Contributor I
edited November 2018 in Help

Hi,

I am new to Rapidminer and am confused about cross validation, optimisation grid, train and testing.

 

I have 2 separate datasets i.e. 1 for training and 1 for testing.  I want to build a SVM model with cross validation and to use optimisation grid to get the best hyperparameters.  My questions are,

 

1.  Do I nest the optimisation grid inside the cross validation operator or do I nest the cross validation operator inside the optimisation grid?

 

2.  Once the optimisation is completed, do I manually get the parameters eg C and gamma values for the SVM model and build separately a SVM model and use an Apply Model & Performance operator with the test data?  Or there is a better way to do this?  A picture of the process flow is much appreciated as I am unable to visualise how it looks.

 

Thanks,

Lobbie

Tagged:

Best Answer

  • Telcontar120Telcontar120 Moderator, RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,625   Unicorn
    Solution Accepted

    You put the cross-validation into your optimization grid operator, and then select the parameters you want to optimize.  The optimization operator will then output the optimized model for you as well as a report of what the selected parameters were.  If you pull the Optimize (Grid) operator into your process pane and then look at the help, there is a link to open a tutorial process (this is true of most operators, by the way), and it will show you a process with the most common configuration.  That should get you started.

     

     

     

     

    Brian T.
    Lindon Ventures 
    Data Science Consulting from Certified RapidMiner Experts

Answers

  • LobbieLobbie Member Posts: 10 Contributor I

    Brian, thank you for your help.  Much appreciated.

Sign In or Register to comment.