Hitachi Vantara Pentaho Community Forums
Results 1 to 3 of 3

Thread: J45 output

  1. #1
    Join Date
    Feb 2016
    Posts
    2

    Default J45 output

    Hello there,
    In the TextArea of the ClassifierPanel we can see a model in the form of a decision tree, when we train J48 algorithm.
    In this decision tree (x/y) is depicted in each leaf of the tree.
    My question is: Is possible to have the same decision tree, but with the values (x/y) representing the results of the classifier after testing the classifier in a test dataset?

    Best regards,
    Adrian
    Last edited by apinoa85; 02-10-2016 at 04:01 AM.

  2. #2
    Join Date
    Aug 2006
    Posts
    1,741

    Default

    Not without hacking the code I'm afraid.

    Cheers,
    Mark.

  3. #3
    Join Date
    Feb 2016
    Posts
    2

    Default

    Quote Originally Posted by Mark View Post
    Not without hacking the code I'm afraid.

    Cheers,
    Mark.
    Hello Mark, Thanks for you fast reply.
    I have experience programming on weka's framework, so if you give me some hints, probably I could do it.
    I realized weka stores the (x/y) values in an object m_tree of the class ClassifierTree. I guess this values are computed
    during the training of the classifier (build(Instance) method).
    In addition, somehow each instance in the test set passed as a parameter to the method classifyInstance(Instance) is run over all the conditions
    in m_tree. I see this might be done in the getProbs(classIndex, instance, weight) method of the class ClassifierTree.
    However, this method is a little confused for me, and I can not find how the test instance is tested in the tree.
    If you could give me additional information, that would be good.
    Thanks in advance,
    Adrian

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.