AutoModel Attribute weights

f2013106f2013106 MemberPosts:1Learner I
edited June 2019 inHelp

How is AutoModel computing attribute weights(which weight by feature operator is it using?) and correlation matrix(is it pearson?)? It seems we can only check the process(through 'open process') for the models and not for feature weights and correlations

Answers

  • yyhuangyyhuang Administrator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, MemberPosts:363RM Data Scientist

    @f2013106You guess is absolutely right here. The weights are normalized from "weight by correlation" and we only keep top 10 most "relavant" features in the auto model view. The correclations uses the Pearson correlation coefficient. We will add more help info in documentations for auto model.

    Thanks for your feedback!

    sgenzer
Sign InorRegisterto comment.