Re:-AUC,ROC,Precision,Recall

guptasha
guptasha New Altair Community Member
edited November 5 in Community Q&A
How to make the charts for AUC, ROC, Precision, Recall for the multi-classification problem? Which operator do I need to use it?
Tagged:

Best Answer

  • varunm1
    varunm1 New Altair Community Member
    edited March 2019 Answer ✓
    Hello @guptasha

    To get an AUC, you need to have a ROC curve. ROC curve is plotted for Binary classification. In the case of multiple class classification, you can get multiple ROC curves when you do one vs all classes type classification. This can be done in rapidminer using Polynomial to Binomial operator. Once you get AUC's for all classes, then you can average them for final AUC of Multiclass.

    Recall and Precision can be found directly with multi-classification, using performance (classification) operator. In this operator, you have weighted mean recall and weighted mean precision options to choose.

    Similar discussion here
    https://community.rapidminer.com/discussion/55258/specificity-sensitivity-and-auc-measures-via-rm-v9-1#latest

    Thanks 

Answers

  • varunm1
    varunm1 New Altair Community Member
    edited March 2019 Answer ✓
    Hello @guptasha

    To get an AUC, you need to have a ROC curve. ROC curve is plotted for Binary classification. In the case of multiple class classification, you can get multiple ROC curves when you do one vs all classes type classification. This can be done in rapidminer using Polynomial to Binomial operator. Once you get AUC's for all classes, then you can average them for final AUC of Multiclass.

    Recall and Precision can be found directly with multi-classification, using performance (classification) operator. In this operator, you have weighted mean recall and weighted mean precision options to choose.

    Similar discussion here
    https://community.rapidminer.com/discussion/55258/specificity-sensitivity-and-auc-measures-via-rm-v9-1#latest

    Thanks 

Categories