The R Book

Libraries for spatial statistics

In addition to the built-in library spatial there are two substantial contributed packages for analysing spatial data. The spatstat library is what you need for the statistical analysis of spatial point patterns (below left), while the spdep library is good for the spatial analysis of data from mapped regions (right).

With point patterns the things you will want to do include

• creation, manipulation and plotting of point patterns,
• exploratory data analysis,
• simulation of point process models,
• parametric model fitting,
• hypothesis tests and diagnostics,

whereas with maps you might

• compute basic spatial statistics such as Moran's I and Geary's C,
• create neighbour objects of class nb,
• create weights list objects of class lw,
• work out neighbour relations from polygons (outlines of regions),
• colour mapped regions on the basis of derived statistics.

You need to take time to master the different formats of the data objects used by the two libraries. You will waste a lot of time if you try to use the functions in these libraries with your own, unreconstructed data files.

Here is the code for installing and reading about spatstat and spdep:

install.packages("spatstat")
library(help=spatstat)
library(spatstat) demo(spatstat)
install.packages("spdep")
library(help=spdep)
library(spdep)