"SVM Operator... How does it Scale Data?"

dragoljubdragoljub MemberPosts:241Maven
edited May 2019 inHelp
Hi Guys,

I am using the built in SVM operator to perform feature weighting. There is a "Scale" check box option that is by default enabled. Which type of scaling is preformed? Range scaling [0 1]? The reason I ask is because the feature weighting differs significantly depending on using or not using the internal scaling.

Regarding feature weighting in general, why does RM not allow weights to be returned from LibSVM? They are stored in the model in the results view? ???


Thanks,
-Gagi
Tagged:

Answers

  • dragoljubdragoljub MemberPosts:241Maven
    Just a friendly bump... ;D
  • landland RapidMiner Certified Analyst, RapidMiner Certified Expert, MemberPosts:2,531Unicorn
    Hi,
    friendly spoken, I ignored this topic yesterday:)Well, if you refer to the SVM Weighting operator I guess it applies a 0-1 weighting. But this should be obvious if you compare the results with the switch on and off.

    To the second question: I don't know:)

    Greetings,
    Sebastian
Sign InorRegisterto comment.