moderncor 0.2.0
New Features
- Added robust correlation methods: biweight midcorrelation,
percentage bend correlation (
WRS2), and Winsorized
correlation (WRS2).
- Added ordinal correlation methods: polychoric and tetrachoric
correlation (
psych).
- Added partial correlation methods: partial correlation and
semi-partial correlation (
ppcor), controlled via the new
z parameter.
- Added nonparametric methods: ball correlation (
Ball)
and Bergsma-Dassios tau* (TauStar).
- Added
moderncor_cat() for categorical association
measures: Cramer’s V, phi coefficient, Goodman-Kruskal gamma, Somers’ D,
contingency coefficient, and Tschuprow’s T (all via
DescTools).
- Added
available_methods_cat() to list all supported
categorical methods.
Changes
- Removed dependency on
WGCNA (Bioconductor). Biweight
midcorrelation is now computed using a built-in implementation of the
standard formula (Wilcox 2012), eliminating the need for
Additional_repositories.
- Version bumped to 0.2.0 to reflect the substantial addition of Phase
2 methods.
moderncor 0.1.0
- Initial release of the
moderncor package.
- Provided a single entry point
moderncor() to compute 9
classical and modern association measures: Pearson, Spearman, Kendall,
distance correlation, MIC, HSIC, Chatterjee’s xi, Hoeffding’s D, and
mutual information.
- Standardized S3 class
"moderncor" with printing,
summary, and as.data.frame methods.
- Supported fast execution by allowing users to toggle p-value
calculations (
p_value = FALSE) to skip slow permutation
tests.
- Implemented utility functions
available_methods() and
method_info() to explore supported algorithms.
- Configured graceful dependency handling for suggested packages via
check_suggested().