MASSExtra (version 1.2.2)

hr_levels: #' @rdname kde_1d #' @export kernelBiweight <- function(x, mean = 0, sd = 1) h <- sqrt(7)*sd ifelse((z <- abs(x-mean)) < h, 15/16*(1 - (z/h)^2)^2/h, 0)

Description

#' @rdname kde_1d #' @export kernelCosine <- function(x, mean = 0, sd = 1) h <- sqrt(1/(1-8/pi^2))*sd ifelse((z <- abs(x-mean)) < h, pi/4*cos((pi*z)/(2*h))/h, 0)

Usage

hr_levels(x, ...)
# S3 method for default
hr_levels(x, p = (1:9)/10, ...)
# S3 method for kde_2d
hr_levels(x, ...)

Value

A vector of density levels defining the home range contours

Arguments

x

an object whose z component represents the KDE

...

extra arguments (currently not used)

p

a vector of probability levels

Details

#' @rdname kde_1d #' @export kernelEpanechnikov <- function(x, mean = 0, sd = 1) h <- sqrt(5)*sd ifelse((z <- abs(x-mean)) < h, 3/4*(1 - (z/h)^2)/h, 0)

#' @rdname kde_1d #' @export kernelGaussian <- function(x, mean = 0, sd = 1) dnorm(x, mean = mean, sd = sd)

#' @rdname kde_1d #' @export kernelLogistic <- function(x, mean = 0, sd = 1) stats::dlogis(x, mean, sqrt(3)/pi*sd)

#' @rdname kde_1d #' @export kernelOptCosine <- function(x, mean = 0, sd = 1) h <- sqrt(1/(1-8/pi^2))*sd ifelse((z <- abs(x-mean)) < h, pi/4*cos((pi*z)/(2*h))/h, 0)

#' @rdname kde_1d #' @export kernelRectangular <- function(x, mean = 0, sd = 1) h <- sqrt(3)*sd ifelse(abs(x-mean) < h, 1/(2*h), 0)

#' @rdname kde_1d #' @export kernelSquaredCosine <- function(x, mean = 0, sd = 1) h <- sqrt(3/(1-6/pi^2))*sd ifelse((z <- abs(x-mean)) < h, cos(pi*z/(2*h))^2/h, 0)

#' @rdname kde_1d #' @export kernelTriangular <- function(x, mean = 0, sd = 1) h <- sqrt(24)*sd/2 ifelse((z <- abs(x-mean)) < h, (1 - z/h)/h, 0)

#' @rdname kde_1d #' @export kernelTricube <- function(x, mean = 0, sd = 1) h <- sqrt(243/35)*sd ifelse((z <- abs(x - mean)) < h, 70/81*(1 - (z/h)^3)^3/h, 0)

#' @rdname kde_1d #' @export kernelTriweight <- function(x, mean = 0, sd = 1) h <- sqrt(9)*sd ifelse((z <- abs(x-mean)) < h, 35/32*(1 - (z/h)^2)^3/h, 0)

#' @rdname kde_1d #' @export kernelUniform <- function(x, mean = 0, sd = 1) h <- sqrt(3)*sd ifelse(abs(x-mean) < h, 1/(2*h), 0)

Home Range levels

For an object representing a 2-dimensional kernel density estimate find the level(s) defining a central "home range" region, that is, a region of probability content p for which all density points within the region are higher than any density point outside the region. This makes it a region of probability p with smallest area.

Examples

krc <- with(Boston, {
  criminality <- log(crim)
  spaciousness <- sqrt(rm)
  kde_2d(criminality, spaciousness)
})
plot(krc, xlab = expression(italic(Criminality)),
          ylab = expression(italic(Spaciousness)))
home <- hr_levels(krc, p = 0.5)
contour(krc, add = TRUE, levels = home, labels = "50%")