• New vignette and link to published article.
  • Replaces the example dataset with one based on real data.
  • Adds confidence intervals to correlation tests and allows different confidence levels.
  • Uses “htest” S3 class for the p-value print method. This is a minor change that makes the output more consistent with other R functions.
  • Corrects a severe error with the p-value computation. Thanks to @ouroboro for pointing this out (#3).
  • Provides an implementation of chebyshev_eval, gammafn, lgammacor and stirlerr functions in C++. This solves CRAN warnings about using non-API non-API calls to R.
  • Provides the option to pass a matrix to obtain a correlation matrix.
  • Uses parallelization to speed up the computation of the Kendall’s correlation.
  • It still needs additional checks with Inf values.
  • The Kendall’s correlation function from capybara was moved here to have a dedicated package for it.