Adaptive Kernel Estimators for Point Process Intensities on Linear Networks

Adaptive estimation of the first-order intensity function of a spatio-temporal point process using kernels and variable bandwidths. The methodology used for estimation is presented in González and Moraga (2022). .

kernstadapt

kernstadapt is an R package for adaptive kernel estimation of the intensity of spatio-temporal point processes.

kernstadapt implements functionalities to estimate the intensity of a spatio-temporal point pattern by kernel smoothing with adaptive bandwidth methodology when each data point has its own bandwidth associated as a function of the crowdedness of the region (in space and time) in which the point is observed.

The package presents the intensity estimation through a direct estimator and the partitioning algorithm methodology presented in González and Moraga (2022).

Installation

The stable version on CRAN can be installed using:

install.packages("kernstadapt")

The development version can be installed using devtools:

# install.packages("devtools") # if not already installed
devtools::install_github("jagm03/kernstadapt")
library(kernstadapt)

Main functions

Direct adaptive estimation of the intensity

• dens.direct() (non-separable)
• dens.direct.sep() (separable)

Adaptive intensity estimation using a partition algorithm

• dens.par() (non-separable)
• dens.par.sep() (separable)

Bandwidths calculation

• bw.abram.temp() (temporal)

Separability test

• separability.test()