R/adkernel.R

In javiergarcirod/stfa: Alternative methods for Stochastic Frontier Analysis

adkernel <-
function (x, kernel = dnorm, bw = bw.nrd0, adjust = 1, n = 500, 
            from, to, cut = 3, na.rm = TRUE,alpha=0.5) 
    
  {
    varname <- deparse(substitute(x))
    if (na.rm) 
      x <- na.omit(x)
    if (!is.numeric(bw)) 
      bw <- bw(x)
    bw <- adjust * bw
    from <- min(x) - cut * bw
    to <- max(x) + cut * bw
    x0 <- seq(from, to, length = n)
    n.1 <- length(x)
    p <- rep(0, n)
    initialp.x0 <- rep(0, n)
    fac <- 1/(n.1 * bw)
    for (i in 1:n) initialp.x0[i] <- fac * sum(kernel((x - x0[i])/bw))
    initialp <- rep(0, n.1)
    for (i in 1:n.1) initialp[i] <- initialp.x0[which.min(abs(x[i] - 
                                                                x0))]
    pbar <- exp((1/n.1) * sum(log(initialp)))
    f <- (initialp/pbar)^-alpha
    for (i in 1:n) p[i] <- fac * sum((1/f) * kernel((x - x0[i])/(f * 
                                                                   bw)))
    result <- list(x = x0, y = p, n = n, bw = bw * adjust, call = match.call(), 
                   data.name = varname, has.na = FALSE)
    class(result) <- "density"
    result
  }