#' Normal Scores
#' 
#' This function applies a quantile-quantile transformation to 
#' the data, resulting in a distribution that is approximately normal
#' but has the same ranks as the original data
#' 
#' @param y A vector. 
#' @param ties.method The option \code{ties.method} in the \code{rank} function.
#' 
#' @return A vector of the same length as \code{y}. 
#' 
#' @author Peter Hoff
#' 
#' @export
#'
#' @examples
#' y<-rexp(100)
#' z<-zscores(y) 
#' par(mfrow=c(1,3))
#' hist(y) 
#' hist(z)
#' plot(y,z) 
zscores<-function(y,ties.method="average")
{
  z<-qnorm(rank(y,na.last="keep",ties.method=ties.method)/(sum(!is.na(y))+1) )

  names(z)<-names(y)

  m<-dim(y)
  if(length(m)==2){ z<-matrix(z,nrow=m[1],ncol=m[2]) ; dimnames(z)<-dimnames(y) }
  if(length(m)>=3){ z<-array(z,dim=m) ; dimnames(z)<-dimnames(y) }

  z
}