R/cov2.R

In RGCCA: Regularized and Sparse Generalized Canonical Correlation Analysis for Multiblock Data

#' Covariance
#'
#' cov2() is similar to stats::cov() but has an additional argument.
#' The denominator \eqn{n} (bias = TRUE) can be used (instead of \eqn{n-1})
#' to give a biased estimator of the (co)variance.
#' Note that if values are missing the "pairwise.complete.obs"
#' option is used resulting in (co)-variance matrix that are not necessarily
#' positive definite.
#' @param x A numeric vector, matrix or data.frame.
#' @param y A numeric vector, matrix or data.frame.
#' @param bias A logical value. If bias = TRUE, \eqn{n} is used to give a
#' biased estimator of the (co)variance.
#' If bias = FALSE, \eqn{n-1} is used.
#' @return Estimation of the variance (resp. covariance) of
#' x (resp. x and y).
#' @title Variance and Covariance (Matrices)
#' @noRd
cov2 <- function(x, y = NULL, bias = TRUE) {
  n <- NROW(x)

  if (is.null(y)) {
    x <- as.matrix(x)
    if (bias) {
      C <- ((n - 1) / n) * stats::cov(x, use = "pairwise.complete.obs")
    } else {
      C <- stats::cov(x, use = "pairwise.complete.obs")
    }
  } else {
    if (bias) {
      C <- ((n - 1) / n) * stats::cov(x, y, use = "pairwise.complete.obs")
    } else {
      C <- stats::cov(x, y, use = "pairwise.complete.obs")
    }
  }
  return(C)
}