A new correlation coefficient

[rflperry]

I've seen some recent discussions surrounding the paper A new coefficient of correlation which basically introduces a nonparametric alternative to the spearman/kendal correlation coefficients which is just as simple but more interpretable. There's also a conditional correlation extension. There's a corresponding analytical null to provide pvalues and the test is asymmetric which is interesting.

They provide an R package but it could be an interesting addition since I assume scipy would require more prevalence/citations. They compare tests on a variety of datasets just like the mgc paper.

[hlabella]

There are some recent updates for a revised variant of Chatterjee rank correlation that "overcome the “only one disadvantage” of Chatterjee’s rank correlation" according to this paper.