R/utils.R

In lookout: Leave One Out Kernel Density Estimates for Outlier Detection

nsphere <- function(n = 3L, r = 1L, npts = 3145L) {
  x <- matrix(1, nrow = npts, ncol = n)
  for (i in seq(n - 1)) {
    j <- n - i
    theta <- seq(from = -pi, to = pi, length.out = npts)
    theta1 <- sample(theta, size = length(theta))
    x[, seq(j)] <- x[, seq(j)] * cos(theta1)
    x[, (j + 1)] <- x[, (j + 1)] * sin(theta1)
  }
  return(x * r)
}

# Unitize each column of X
unitize <- function(X) {
  for (col in 1:NCOL(X)) {
    maxcol <- max(X[, col])
    mincol <- min(X[, col])
    if(maxcol!= mincol){
      X[, col] <- (X[, col] - mincol) / (maxcol - mincol)
    }
  }
  X
}