Options

"Support Vector Machine (Evolutionary) only predicting true outcomes"

Eric005Eric005 Member Posts: 9 Contributor I
edited June 2019 in Help

Just curious why this is the case, using a SVM on a sliding window time series validation for a binary indication and the SVM is only providing a true estimate.  Is this becuase the SVM estimates the 1 of 2 cases (true or not true) instead of postively true or postively false?  Is there a workaround?  Thanks!

 

 

Tagged:

Answers

  • Options
    MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,517 RM Data Scientist

    Hey,

     

    are you sure that your windows are not sometimes completly one class or highly skewed?

     

    Best,

    Martin

    - Sr. Director Data Solutions, Altair RapidMiner -
    Dortmund, Germany
  • Options
    Eric005Eric005 Member Posts: 9 Contributor I
    Thank you Martin. The data being analyzed is a set of fairly high-frequency market data over a number of periods so it's nearly impossible to conceive of a situation where it would be directionally skewed so uniformly. Any thoughts?
Sign In or Register to comment.