R/getnorm.naive.R

In TULIP: A Toolbox for Linear Discriminant Analysis with Penalties

getnorm.naive <- function(x, y){
  f0 <- ecdf(x[y==1])
  n0 <- sum(y==1)
  n1 <- sum(y==2)
  n <- n0+n1
  delta.n0 <- 1/n0^2
  
  v0 <- rep(0,n)
  v0[y==1] <- qnorm(truncate(x[y==1],f0,delta.n0))
  v0[y==2] <- qnorm(truncate(x[y==2],f0,delta.n0))
  mu0.hat <- mean(v0[y==2])
  
  
  transform <- function(t){
    qnorm(truncate(t,f0,delta.n0))  
  }
  x.norm <- transform(x)
  
  list(x.norm=x.norm, mu.hat=mu0.hat, transform=transform)
}