Barry Rowlingson, Adrian Baddeley, Rolf Turner, Peter Diggle
Rasp: A Package for Spatial Statistics
************************************************************

This paper describes the development of a new R package for spatial
data and statistics. We review existing packages that deal with
spatial information and comment on how they handle the basic data. Our
goal is then to construct a unifying spatial data structure that can
form a base for any spatial statistics packages.

We build on our experience with earlier packages -- spatstat
(http://www.maths.uwa.edu.au/~adrian/spatstat.html) and
splancs (http://www.maths.lancs.ac.uk/Software/Splancs/) -- to
produce a package with extensive data-manipulation and statistical
functionality. An object-oriented design is used throughout to
abstract functionality from any implementation specifics, but we take
care to present the user with simple functions that have sensible
defaults. The paper is as much a technical document on package design
as it is a story of how such a package evolves from the ideas of the
authors.

The basic functionality deals with coordinates and associated
attributes of each coordinate. The coordinates may be Cartesian spatial
locations in up to three dimensions, together with an optional
temporal location to handle space-time data. Data structures and
methods for polygonal regions are also implemented.

Currently we have functions for spatial point-pattern analysis derived
from spatstat and splancs. The geostatistical functionality of the geoR
package can be easily integrated, as well as other packages currently
in the CRAN.