Correlation between polynominal attributes?

rapid1234rapid1234 Member Posts: 5 Learner I
Hi, I work with the employee attrition data set and have a question. The data set contains numerical values, bionominal and polynomial values. With the nummerical and bionominal values, I can use the operator correlationmatrix and see how the dependency on the target variable attribution is. How do I do that with polynomial values like business travel? Could I use the operater nominal to nummerical? or are there better alternatives to show the dependencies between the label and the polynomial values?

Thank you!
Tagged:

Answers

  • MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,503 RM Data Scientist
    Correlation is not defined for non-numerical data types. you need to use a measure which works on nominal data For example gini index.
    - Sr. Director Data Solutions, Altair RapidMiner -
    Dortmund, Germany
  • rapid1234rapid1234 Member Posts: 5 Learner I
    @mschmitzI have to choose "normalized weight" at gini index? Otherwise the correlation is very low.
  • MartinLiebigMartinLiebig Administrator, Moderator, Employee, RapidMiner Certified Analyst, RapidMiner Certified Expert, University Professor Posts: 3,503 RM Data Scientist
    Usually not. Normalize weight normalizes all values in a way, that the top value is 1.

    Keep in mind, that the normalization of gini index is totally different to the normalization of correlation. A gini index of 0.1 is usually a powerful attribute.

    BR,
    Martin
    - Sr. Director Data Solutions, Altair RapidMiner -
    Dortmund, Germany
Sign In or Register to comment.