qkdeSorted: A slightly altered version of qkde from package ks
In MaskJointDensity: Masking, Unmasking and Restoring Confidential Data

Description Usage Arguments Details Value Author(s) References Examples

Qkde complains if the output is not sorted. Theoretically it is impossible for this to happen, but in practise floating point error can allow this to occur. This function is a cheap work-around to this problem.

1	qkdeSorted(p, fhat)

`p`	See qkde in package ks
`fhat`	See qkde in package ks

See qkde in package ks

Luke Mazur

see package ks

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (p, fhat) 
{
    if (any(p > 1) | any(p < 0)) 
        stop("p must be <= 1 and >= 0")
    cumul.prob <- pkde(q = fhat$eval.points, fhat = fhat)
    ind <- findIntervalSorted(x = p, vec = cumul.prob)
    quant <- rep(0, length(ind))
    for (j in 1:length(ind)) {
        i <- ind[j]
        if (i == 0) 
            quant[j] <- fhat$eval.points[1]
        else if (i >= length(fhat$eval.points)) 
            quant[j] <- fhat$eval.points[length(fhat$eval.points)]
        else {
            quant1 <- fhat$eval.points[i]
            quant2 <- fhat$eval.points[i + 1]
            prob1 <- cumul.prob[i]
            prob2 <- cumul.prob[i + 1]
            alpha <- (p[j] - prob2)/(prob1 - prob2)
            quant[j] <- quant1 * alpha + quant2 * (1 - alpha)
        }
    }
    return(quant)
  }