"How does correlation matrix operator calculate correlation of nominal values?"

njasajnjasaj Member Posts: 18 Contributor II
edited June 2019 in Help
I need to know that how does correlation matrix operator calculate correlation between a continues variable and a discrete variable? does it use Pearson correlation coefficient?


  • Options
    MariusHelfMariusHelf RapidMiner Certified Expert, Member Posts: 1,869 Unicorn

    in RapidMiner, discrete variables are mapped internally to an integer value. When calculating the correlation matrix, that integer mapping is used to calculate the pearson correlation coefficient.

    But since the mapping is somewhat non-deterministic and not the same for different datasets with the same discrete values, it is not valid to rely on the coefficients calculated for nominal values by the operator. You should rather convert the nominal values to numerical values with one of the various operators made for that, e.g. Nominal to Numerical.

    Best regards,
Sign In or Register to comment.