How can I change type of weight in feature selection algorithms?

maryam_raz21
maryam_raz21 New Altair Community Member
edited November 2024 in Community Q&A

Hi. I'm new in rapidminer and I try feature selection algorithm such as backward eliminatiom and svm-RFE. The problem is that in the result the weight of feature is only 0,1 and it's not useful for me. How can I chang it to real?

Tagged:

Answers

  • kypexin
    kypexin New Altair Community Member

    Hi @maryam_raz21

     

    To get real weights, you should use one of the 'Weight by ...' operators; the feature selection operators you mentioned are acutually only selecting relevant features (hence assigning them weight = 1) and drop irrelevant (hence assigning them weight = 0).

  • maryam_raz21
    maryam_raz21 New Altair Community Member

    Thanks for reply:smileyhappy: . I have  another question . This feature selection such as SVM-RFE always selecting relevant or just in rapidminer They behave like this؟ in the papers that I studeid, they used svm-RFE for bioinformatics widely!

  • MartinLiebig
    MartinLiebig
    Altair Employee

    Hi,

     

    have you checked the feature selection extension for this?

     

    Cheers,

    Martin

  • maryam_raz21
    maryam_raz21 New Altair Community Member

    yes. I installed feature selection extension 1.1.4 that included svm-rfe, ensemble feature selection and etc. in result all of feature's weight that were selected are 1.:smileyindifferent:

Categories