independence

Performs three ranking-based nonparametric tests for the independence of two continuous variables: (1) the classical Hoeffding's D test; (2) a refined variant of it, named R; (3) the Bergsma-Dassios T* sign covariance. The first test is consistent assuming an absolutely continuous bivariate distribution, i.e., the population coefficient D=0 iff the variables are independent. The latter two are consistent under no restriction on the distribution. All three statistics are computed in time O(n log n) given n iid paired samples. The computation of R and T* uses a new algorithm, following work of Even-Zohar and Leng (2019), see , .