Clarifications Needed on Validation and Performance Output

deiegit2k4

Hi all, I am new here and have some questions which I need answers to.

1. Which one is better for classification between split validation and cross validation?

2. How to use/introduce split data (manually) while using cross validation?

3. How to display the test results (performance output) of a model showing all the the predicted attribute while using optimization as well as cross validation?

 

Thanks,

Mekks

sgenzer

Hello @deiegit2k4 - welcome to the community.  I'm moving this thread to "Getting Started" as I think you will see more traction there.  FYI there are MANY threads about these same questions here on the community.  Have you tried "Search the community..."?

 

Scott

 

Telcontar120

You should check out Ingo's excellent video series and blog posts on model validation, all available from the RapidMiner site.  The "Getting Started Central" page also has video tutorials that show how to do both cross-validation and model optimization as well.  The short answer is that cross-validation is superior to split validation in almost every way and should be considered best practice.  You can use both together (just use the "Split Data" operator before doing cross validation), but separate split validation isn't really necessary if you are doing cross-validation.