a newbie question about testing model

amenaakhterchyamenaakhterchy Member Posts: 7 Contributor I
edited November 2018 in Help

hi i m trying to test a dataset .. i divided 1 dataset into 2 part 80% & 20 % .. now i created a model with 80% data ..my question is i want to test 20% dataset with the same model ..how can i do this?

Answers

  • Telcontar120Telcontar120 Moderator, RapidMiner Certified Analyst, RapidMiner Certified Expert, Member Posts: 1,635 Unicorn

    I recommend using the "split validation" operator, it will basically do things for you once you specify the split you want to use.  You'll need a modeling operator on the left and apply model and a performance operator on the right.  The built-in tutorial will show you how to do it.

     

    You can of course do this manually as well, using the split operator, and then doing the apply model and performance operators.

     

    Remember that you'll need to apply the same set of preprocessing steps to both the training and the testing dataset.  Apply models can be used to apply multiple preprocessing models such as normalization or other transformations.

     

     

    Brian T.
    Lindon Ventures 
    Data Science Consulting from Certified RapidMiner Experts
  • IngoRMIngoRM Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, Community Manager, RMResearcher, Member, University Professor Posts: 1,751 RM Founder

    Brian's answer is perfect.  I ust wanted to add that I recently started a new blog series on correct model validation which you can find - surprise! - on our blog.  Here is the link to the first part:

     

    https://rapidminer.com/validate-models-training-test-error/

     

    Parts 2 and 3 are also published by now.

     

    Hope this helps,

    Ingo

Sign In or Register to comment.