For statistical analysis, the most complete computing environment available is the R programming language. The success of R depends on the highly active community of developers of the add-on packages which facilitate a vast range of cutting-edge data analyses. My contributions include these add-on R packages for multivariate data analysis, available on CRAN (Comprehensive R Archive Network).

R packages

ks Kernel Smoothing

Flagship product for non-parametric kernel data smoothing with R base graphics for 1- to 6-dimensional data. This package implements a wide range of multivariate kernel smoothers (e.g. density estimation, density derivative estimation, classification, clustering, regression etc.), as well as sophisticated visualisation graphics. It also forms the basis for the practical data analysis contained in the book Multivariate Kernel Smoothing and Its Applications.

eks Tidy and Geospatial Kernel Smoothing

Extension of the flagship package ks package for ggplot2 graphics (1- and 2-dimensional data) and geospatial 2-dimensional data. This package facilitates access for tidyverse users to the wide suite of kernel smoothers in ks. In addition, it is compatible with geospatial data coded as simple features from the sf package, for both base R and ggplot2 graphics.

feature Local Inferential Feature Significance for Multivariate Kernel Density Estimation

Extension of the ks package to feature significance for 1- to 3-dimensional data.

prim Patient Rule Induction Method (PRIM) for high-dimensional data

Bump hunting and highest density difference region estimation for high-dimensional data.

Other R sofware

Other software that I've developed in R, but haven't been uploaded to CRAN as packages, include:

MARTS (Markov Assignment for Road Traffic Systems)

Code for stochastic assignment in traffic networks.

Interface for R and LinBugs

Code for calling LinBugs inside R on a Unix-type system.