Different performance between grid search and real test

YADAOYADAO MemberPosts:1Newbie
edited October 2020 inHelp
Hi guys,
I used optimize parameters to find the best parameters' values with 10 fold cross-validation, and it showed out that the best kappa was 0.397. But when I applied these parameter values to 10 fold cross-validation myself, the kappa was only 0.337.
What's wrong?
Thank you.
Tagged:

Answers

  • MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University ProfessorPosts:3,368RM Data Scientist
    Hi@YADAO,
    can it just be sampling differences? What happens if you change the seed of your x-val?

    Best,
    Martin
    - Head of Data Science Services at RapidMiner -
    Dortmund, Germany
  • swordfswordf MemberPosts:3Contributor I

    Is there still someone on this issue now?
    I have the same problem, different results are presented using the suggested parameter by grid optimizer,

    I don't understand why, even if it's not the same it should be close at least
  • MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University ProfessorPosts:3,368RM Data Scientist
    close in number of standard_deviations, yes.
    Best,
    Martin
    - Head of Data Science Services at RapidMiner -
    Dortmund, Germany
Sign InorRegisterto comment.