Measurement Level Independent Feature Correlation Matrix

Uses three different correlation coefficients to calculate measurement-level adequate correlations in a feature matrix: Pearson product-moment correlation coefficient, Intraclass correlation and Cramer's V.

featureMatrix

R package Feature Correlation Matrix

Status: First version (0.4.0) ready to get pushed to CRAN, minor changes under development already !

Authors: Dr. Guido Möser/ Ilja Muhl (in alphabetical order)

Notes: If you want to run the featureMatrix()-function, the two other functions, cv.test() and icc(), have to be available!

What it does: Estimates the correlation between categorical and numerical features (and the target variable)

Parameter:

• both variables numerical: Product-Moment-Correlation by Bravais-Pearson
• both variables categorical: Cramer's V (based on the Chi²-Test of Independence)
• one variable categorical and the other numerical: ICC (Intra-Class-Correlation) based on classical ANOVA by Fisher