R/rescale.R

In cbdAmgen/bcgp0a: Bayesian Composite Gaussian Processes

#' Rescale a vector to [0, 1]
#' @description Rescales a vector to [0, 1]
#' @details Rescales a vector to [0, 1].
#' @usage rescale(x)
#' @param \code{x} A vector input
#' @return A vector rescaled to [0, 1]
#' @examples
#' rescale(rnorm(10, 0, 100))
#' @author Casey Davis (\email{cbdavis33@@gmail.com})
rescale <- function(x){

  minX <- min(x)
  rangeX <- max(x) - minX
  scaled <- (x - minX)/rangeX

  return(scaled)

}


#' Scale a matrix to the unit hypercube
#' @description Scales a matrix to \deqn{[0, 1]^d}
#' @details This function scales a matrix to the unit hypercube by subtracting off
#' the minimum value in each column from each value and dividing by the range of
#' that column.
#' @usage scaleX(x)
#' @param \code{x} An \code{n x d} matrix
#' @return A matrix rescaled to \deqn{[0, 1]^d} with attributes
#' \code{"scaled:minimum"} and \code{"scaled:range"}
#' @examples
#' scaleX(matrix(runif(20, 0, 10), nrow = 10, ncol = 2))
#' @author Casey Davis (\email{cbdavis33@@gmail.com})
scaleX <- function(x){

  xScaled <- apply(x, 2, rescale)
  rangeX <- apply(x, 2, range)
  xScaled <- structure(xScaled,
                       'scaled:minimum' = rangeX[1,],
                       'scaled:range' = rangeX[2,] - rangeX[1,])
  return(xScaled)
}