Smart Spatial Analysis
Introduction
These pages document developments in the ESRC funded project entitled "A smart
spatial pattern explorer for the geographical analysis of GIS data", the
principal investigator is
Prof. Stan Openshaw, with Dr Ian
Turton as researcher.
Background
The galloping computerisation of nearly all of society's administrative
and management functions is creating an immensely data rich
situation. Much of these data are geographically referenced; for
instance by postal address or postcode. Unfortunately, there are as yet
few geographical analysis methods able to explore large databases for
evidence of patterns if the analyst has no good ideas
of where and when to look for the patterns and what characteristics they
may have. As a result many important datasets in both the public and
private sector are not being analysed to the fullest extent, if indeed
at all. For example, data about crime, disease and marketing behaviour
are collected and at best poorly analysed using often inadequate
methods.
Aims
This site aims to develop a novel new approach to the
exploratory analysis of geographical data that is widely applicable. The
idea is to use techniques developed in an area of artificial
intelligence known as artificial life to create artificial pattern
hunting creatures able to move around the complex many dimensional space
of a complex multivariate databases in a search for potential patterns
where further investigation may be worthwhile. The aim is to develop an
automated, smart geographical analysis tool that will make suggestions
as to where to look for patterns, when
to look, and what to look for. This is not an easy task
but the problems can be resolved by the use of computationally intensive
methods based upon a supercomputer. The dynamics of these pattern
hunting creatures will be visualised using computer animation so that
the end-user watching these computer movies will be able to discover
whether the patterns being uncovered are real or figments of the
imagination. It is believed that perfecting this automated method of
analysis will have a most significant impact on the usefulness of many
large geographically referenced databases in many different application
areas.
Results
This project came on-line in October 1997, as various systems are
developed and tested they will be listed below.
GAM/K
The original GAM/1 geographical analysis machine
was developed by Openshaw et al (1987, 1988).
GAM/K is a development by Openshaw and Craft (1991) this version
has now been recoded and a WWW interface added. Users are invited
to test the on-line version of GAM/K.
GEM
The Geographical Exploration Machine (GEM) has now been
converted to work on the WWW in a similar way to GAM above. Again users
are invited to try out
the on-line version of GEM.
Test Data
The CCG has developed two sets of synthetic data
to allow people to test cluster detection methods. A
brief introduction
and the data sets are provided.
References
Openshaw, S., Charlton, M., Wymer, C., and Craft, A., (1987) 'A mark I
geographical analysis machine for the automated analysis of point data
sets', International Journal of Geographical Information Systems, 1,
p335-358.
Openshaw, S., Charlton, M., Craft, A. and Birch, J. (1988) 'An
investigation of leukaemia clusters by the use of a geographical analysis
machine', The Lancet, Feb 6th, 272-273.
Openshaw, S., and Craft, A., (1991) 'Using geographical analysis machines
to search for evidence of cluster and clustering in childhood leukaemia and
non-Hodgkin Lymphomas in Britain. In G. Draper (ed) 'The Geographical
Epidemiology of Childhood Leukaemia and non-Hodgkin Lymphomas in Great Britain
1966-83' Studies in Medical and Population Subjects No 53, OPCS, London,
HMSO.
Ian Turton < ian@geog.leeds.ac.uk >
Last modified: Mon May 18 13:30:14 BST 1998