Point pattern analysis refers to the examination of patterns of (possibly marked) points to try to establish whether there are regularities in the process they represent (are they clustered, randomly spread or regular?). A key feature is how the region of interest is defined, and how edge effects are accommodated. It is also important to analyse spatial inhomogeneity.

Spatial data analysis
R spatial projects

Packages on CRAN
Maps
Point pattern
Geostatistics
Areal
GIS interfaces
Others

Other Packages
Details

Mailing list
R-sig-Geo

Links
Back to SAL

spatial

The VR bundle includes the spatial package, which has recommended status, and is thus available in every R installation. Regions of interest are defined as rectangles. This version of the Khat measure should be taken as the reference implementation, since it is directly derived from original research by Brian Ripley. The functions are documented in: William N. Venables and Brian D. Ripley. Modern Applied Statistics with S. Fourth Edition. Springer, 2002.

splancs

splancs has been available for S for some time, and is documented in: Rowlingson, B. and Diggle, P. 1993 Splancs: spatial point pattern analysis code in S-Plus. Computers and Geosciences, 19, 627-655; its homepage provides some further details. The R port adds datasets from: Bailey, T. C. and Gatrell, A. C. (1995) Interactive spatial data analysis, Harlow: Longman, allowing most of the exercises from the point pattern chapters to be run in R. Splancs allows the region of interest to be an arbitrary polygon. The package is the only one available that handles space-time point patterns as well as spatial patterns.

spatstat

spatstat is a newer package, with a homepage. It allows further freedom in defining the region(s) of interest, and makes further extensions to marked processes and spatial covariates. Its strengths are model-fitting and simulation. It is the only package that will enable the user to fit inhomogeneous point process models with interpoint interactions. The fitted model can be simulated automatically. Its comprehensiveness is shown by the large number of functions available. It is by far the largest of the three packages. The package is presented in depth in a paper published in the Journal of Statistical Software, 2005: Spatstat: an R package for analyzing spatial point patterns.

Dylan Beaudette has found that spatstat can be built on PowerPC 64 bit systems by:

export CFLAGS="-mpowerpc64 -O2"

and equivalently for CXXFLAGS, FFLAGS and FCFLAGS. There may be other ways around this, but these do work currently

MarkedPointProcess

The MarkedPointProcess package for the analysis of marks of marked point processes uses the RandomFields package in the analysis of marked point patterns.

spatialkernel

The spatialkernel package for two kinds of kernel smoothing methods, kernel regression estimate of the type-specific probabilities in a multivariate Poisson point process and kernel density estimate of the intensity function of an inhomogeneous Poisson point process with edge-correction algorithm implemented against an arbitrary polygon area is now on CRAN; its website contains further details and references.

spatclus

Multiple cluster location and detection for 2D and 3D spatial point patterns (case event data). The methodology of the spatclus package is based on an original method that allows the detection of multiple clusters of any shape. A selection order and the distance from its nearest neighbour once pre-selected points have been taken into account are attributed at each point. This distance is weighted by the expected distance under the uniform distribution hypothesis. Potential clusters are located by modelling the multiple structural change of the distances on the selection order. Their presence is tested using the double maximum test and a Monte Carlo procedure.