Hitachi Vantara Pentaho Community Forums
Results 1 to 4 of 4

Thread: Interpretation of cfssubseteval with Genetic Search in WEKA

  1. #1
    Join Date
    Feb 2016
    Posts
    5

    Lightbulb Interpretation of cfssubseteval with Genetic Search in WEKA

    Dear All,

    I have carried out feature selection on my training set using genetic search with the evaluator cfsSubsetEval. However i am a little confused with the output from the generations in WEKA and how this ties in with the final subset selected and outputted in WEKA.

    Using this procedure i obtain 65 features. However none of these features in this combination are present in my last generation (20 populations). So i am wondering how WEKA got to these 65 features when the cfssubset evaluates a subset as a whole and not individual features? Surely it would select the subset in my final generation with the highest merit?

    Any help would be much appreciated!

    Regards
    Danielle
    Attached Files Attached Files
    Last edited by dnewby; 02-10-2016 at 10:34 AM.

  2. #2
    Join Date
    Feb 2016
    Posts
    5

    Default

    Hi everyone,
    I have worked this out i think! There is a setting on cfsSubsetEval called locally predictive which default is TRUE. This parameter allows additional features to be added to the best subset from the genetic search. The definition is: “Identify locally predictive attributes. Iteratively adds attributes with the highest correlation with the class as long as there is not already an attribute in the subset that has a higher correlation with the attribute in question”.

  3. #3
    Join Date
    Aug 2006
    Posts
    1,741

    Default

    Yes, this is probably the case. It could also be the case that the best subset was found in an earlier generation. The search continues for max generations and retains the best subset found at any time during the search (not just in the last generation).

    Cheers,
    Mark.

  4. #4
    Join Date
    Feb 2016
    Posts
    5

    Default

    Thanks Mark!
    Quote Originally Posted by Mark View Post
    Yes, this is probably the case. It could also be the case that the best subset was found in an earlier generation. The search continues for max generations and retains the best subset found at any time during the search (not just in the last generation).

    Cheers,
    Mark.

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
Privacy Policy | Legal Notices | Safe Harbor Privacy Policy

Copyright © 2005 - 2019 Hitachi Vantara Corporation. All Rights Reserved.