R/abs_res.R

In generalCorr: Generalized Correlations, Causal Paths and Portfolio Selection

#' Absolute residuals of kernel regression of x on y.
#'
#' This internal function calls the \code{kern} function to implement kernel regression
#' with the option \code{residuals=TRUE} and returns absolute residuals.
#'
#' The first argument is assumed to be the dependent variable.  If
#' \code{abs_res(x,y)} is used, you are regressing x on y (not the usual y on
#' x)
#'
#' @param x {vector of data on the dependent variable}
#' @param y {vector of data on the regressor}
#' @return absolute values of kernel regression residuals are returned.
#' @note This function is intended for internal use.
#' @author Prof. H. D. Vinod, Economics Dept., Fordham University, NY
## @seealso %% ~~objects to See Also as \code{\link{help}}, ~~~ @references %%
## ~put references to the literature/web site here ~
#' @concept  kernel regression residuals
#' @examples
#' \dontrun{
#' set.seed(330)
#' x=sample(20:50)
#' y=sample(20:50)
#' abs_res(x,y)
#' }
#' @export

abs_res <- function(x, y) {
    kk1 = kern(dep.y = x, reg.x = y, residuals = TRUE)
    ares = abs(kk1$resid)
    return(ares)
}