R/aitchisonVariation.R

In pknight24/KPR: Kernel-penalized regression

#' Aitchison's Variation Matrix
#'
#' Computes Aitchison's variation matrix from a matrix X
#' @param X An n x p data matrix.
#' @return A p x p Aitchison variation matrix.
#' @importFrom stats var
#' @export
aitchisonVariation <- function(X)
{
  p <- ncol(X)
  A <- matrix(nrow = p, ncol = p)
  for(i in 1:p){
    for(j in 1:p){
      A[i, j] = var(log(X[, i] / X[, j])) / sqrt(2)  #
    }
  }
  return(A)
}