R/ROCx.r

In logcondens: Estimate a Log-Concave Probability Density from Iid Observations

ROCx <- function(x, res0, res1, smooth = FALSE) {
  qs0 <- quantilesLogConDens(1 - x, res0)

  if (identical(smooth, FALSE)) {
    c <- 3
  }
  if (identical(smooth, TRUE)) {
    c <- 5
  }

  F1 <- evaluateLogConDens(qs0[, "quantile"], res1, which = c, gam = NULL, print = FALSE)

  if (identical(smooth, FALSE)) {
    roc.x <- 1 - F1[, "CDF"]
  }
  if (identical(smooth, TRUE)) {
    roc.x <- 1 - F1[, "smooth.CDF"]
  }

  return(as.numeric(roc.x))
}