kde_2d: Two Dimensional Kernel Density Estimate

In BillVenables/kdeR: Kernel Density Estimates

Description

A pure R implementation of two direct algorithms.

Usage

kde_2d(
  x,
  y,
  N = 256,
  bw = c(x = bw.nrd0(x), y = bw.nrd0(y)),
  lower_x = limits$x$lower,
  upper_x = limits$x$upper,
  lower_y = limits$y$lower,
  upper_y = limits$y$upper,
  k = 1
)

## S3 method for class 'kde_2d'
print(x, ...)

## S3 method for class 'kde_2d'
plot(
  x,
  ...,
  xlab = bquote(italic(.(x$data_name[["x"]]))),
  ylab = bquote(italic(.(x$data_name[["y"]]))),
  col = hcl.colors(25, rev = TRUE)
)

## S3 method for class 'kde_2d'
as.data.frame(
  x,
  row.names,
  optional,
  ...,
  borders = FALSE,
  responseName = "density"
)

Arguments

x	Numeric vector or, in generics, a kde object
y	Numeric vector of the response Variables
N	Integer: number of equally spaced points for the kde. May be one or two for x and y respectively
bw	Numeric: bandwidth(s) for x and y kernels respectively. A single value will be duplicated.
lower_x, upper_x, lower_y, upper_y	Numeric: kde range limits
k	Ineger: number of bandwidths by which to increase the range for the default limits.
...	Extra arguments passed on to methods.
xlab, ylab, col	base graphics parameters
row.names, optional	As for the as.data.frame generic
borders	logical: should zero borders be included around the density when the object is coerced to a data frame.
responseName	character string: name for the density vector.

Value

An object of class "kde_2d"

Examples

set.seed(1234)
x_value <- abs(rnorm(5000))
y_value <- rnorm(x_value)
kxy <- kde_2d(x_value, y_value, lower_x = 0)
kxy
if("package:ggplot2" %in% search()) {
    ggplot(within(as.data.frame(kxy),
                  pDens <- 2*dnorm(x_value)*dnorm(y_value))) +
      aes(x = x_value, y = y_value) +
      geom_raster(aes(fill = density)) +
      geom_contour(aes(z = pDens), colour = "black") +
      scale_fill_viridis_c(direction = -1) +
      theme_bw() + theme(legend.position = "none")
} else {
  with(unclass(kxy), {
    oldPar <- par(mar = c(3, 3, 1, 1) + 0.1, las = 1, mgp = c(1.5, 0.5, 0),
                 cex.axis = 0.7, cex.lab = 0.8, tck = -0.015)
    plot(kxy)
    contour(x, y, z = outer(x, y, function(x, y) 2*dnorm(x)*dnorm(y)),
            nlevels = 6, add = TRUE)
    box()
    par(oldPar)
  })
}

BillVenables/kdeR documentation built on April 12, 2021, 1:51 p.m.