"Decision Tree Optimization and Accuracy"

lex_lex_ Member Posts: 2 Contributor I
edited June 2019 in Help

Hi,

 

I'm trying to use a decision tree which is nested inside the Optimize Parameters (Grid), focusing on Max. Depth and Min. Gain.

 

 

Reconstructing the decision tree using the results obtained above, but without the Optimize Parameters (Grid), the accuracy is lower now.

 

Why is that so?

Best Answer

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

    If you are using cross-validation without a local random seed set then every time you close and re-open that process in RapidMiner you can get a different result even if you don't make any other changes.  So I suspect that could be the issue.  

     

     

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

Answers

  • MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,503 RM Data Scientist

    Hi,

     

    what performances are you comparing? The X-Val performances from within the optimization with an X-Val result from without?


    Are the performances comparable w.r.t their std_devs?


    ~Martin

    - Sr. Director Data Solutions, Altair RapidMiner -
    Dortmund, Germany
  • lex_lex_ Member Posts: 2 Contributor I

    Thanks for pointing me to the right direction.

     

    The random seed was the issue, resulting in the data sets being different after the split operator.

Sign In or Register to comment.