Methods for Automating the Exploration of the Geocyberspace of Health and Crime Data Reviving the Geographical Analysis Machine
Contents
Background
There is a VAST and RAPIDLY growing GEOCYBERSPACE of information
Many Spatial Data Bases are now available for analysis
The NEED is for Exploratory Spatial Data Analysis
Many users now have a requirement for DATA EFFICIENT and USER-FRIENDLY exploratory spatial data analysis methods capable of being safely used by people who do not have higher degrees in the statistical or spatial sciences
User Friendly Spatial Analysis:
These Methods can be classified as follows
PPT Slide
Mark 1 Geographical Analysis Machine
Geographical Analysis Machine (GAM) Mark 1 history
10 years ago GAM/1 was a mixed blessing!
GAM/1: good aspects
GAM/1: Bads
and
it was developed by a GEOGRAPHER
Some of the problems went away
International Agency for Research on Cancer (IARC)
Results published in Alexander and Boyle (1996)
An inverse relationship between the strength of GAM criticism and the performance of the preferred methods!
Clustering Found in 10 Random Data Sets
Detection of Cluster Locations
Finding Correct Clusters
Estimated Positive Sensitivities
Alexander and Boyle (1996) authors of the IARC study concluded:
That was in 1991!!!!!!!!
Reviving GAM/K
Original GAM/K Cray X-MP code
Algorithm was re-programmed from scratch
Making GAM/K run faster
Linking GAM/K to any GIS is easy
How does GAM/K work?
Basic GAM/K algorithm: Part 1
Basic GAM/K algorithm: Part 2
Get Data for Circle (CX,CY,radius)
A more efficient circle retrieval algorithm
Burglary Data for Sheffield
Crime Data is considered to be far easier to analyse that rare disease data
What significance test?
Results: uses Poisson Probabilities
Machine used?
What about multiple testing?
Conclusions
GAM Strengths
GAM Deficiencies
The other GAMs
Future Plans
Further Info: Email
Email: stan@geog.leeds.ac.uk
Home Page: http://www.geog.leeds.ac.uk/staff/s.openshaw