class distribution reversing

NonaNona Member Posts: 15 Contributor II
hi all
I am working on an imbalanced dataset with 2 classes. class 0 has 74 samples in training data and class 1 has 4265 samples. I placed a 'breakpoint after' the training data, and check its stats and it shows class 0-74 and class 1-4265, which is correct. but soon after this, when i press the continue execution (green arrow) it changes to class 1-74 and class 0- 4265 before running the algorithm. To double check this, after the execution was complete, i right clicked on the training set and selected the 'show example set result' option, the stats now show class 1-74 and class 0- 4265, which is incorrect. what is wrong here? i am using remap binomials before performance evaluation of the test set and setting class 0 as positive class. please reply soon.
Thanks

Answers

  • MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,503 RM Data Scientist
    dont rely on the "show example set" after you completed the process. Trust on the breakpoint one.

    If you really want to check it in depth, try to add a store operator after the point you want to ahve it. You can use %{a} to get the current execution count of the operator to store each chunk individually.

    ~Martin
    - Sr. Director Data Solutions, Altair RapidMiner -
    Dortmund, Germany
  • NonaNona Member Posts: 15 Contributor II
    Thank you Martin.
Sign In or Register to comment.