R/normalize.R

In dougmcneall/emtools: What the Package Does (One Line, Title Case)

#' Normalize columns of a matrix onto a [0-1] basis
#'
#' @param X A design matrix
#' @param wrt A design matrix which acts as a reference to normalize X against
#'
#' @return A matrix normalized to [0-1]
#' @export
#'
#' @examples normalize(X)
normalize <- function(X, wrt = NULL){
  # Normalize a matrix to [0,1] on a per-column basis.
  # Normalize relative to matrix wrt if included.

  f <- function(X){
    (X-min(X))/(max(X)-min(X))
  }

  # test to see if we have a matrix, array or data frame
  if(length(dim(X))==2){
    out <- apply(X,2,f)
  }

  else{
    out <- f(X)
  }

  if(is.null(wrt) == FALSE){
    # if argument wrt is given

    n <- nrow(X)
    mmins <- t(kronecker(apply(wrt,2,min),t(rep(1,n))))
    mmaxs <- t(kronecker(apply(wrt,2,max),t(rep(1,n))))

    out <- (X-mmins)/(mmaxs-mmins)
  }

  out
}