Kernel Estimate Intensity at Data Points

Usage

intensity_at_points(
  x,
  bw,
  kernel = "gaussian",
  border = "local",
  normalise = FALSE,
  loo = FALSE
)

Arguments

x: point pattern
bw: bandwidth. Gaussian sd=bw, Epanechnicov domain [-bw, bw].
kernel: Either 'gaussian' or 'epanenchnikov' (partial matching)
border: Border correction to apply. One of 'none', 'local', 'global', 'toroidal'.
normalise: renormalise so that inverse sum = volume
loo: leave-one-out -estimate? Don't include point i for estimation of int(i)

Details

For border correction, the bounding box of the coordinates of x will be used, so at the moment it works only for axis-aligned cuboids.

The kernel will be a product of one dimensional kernels, so Epanechnikov in many dimensions will not be truly isotropic but prefers diagonal directions.