Transcript Advanced Crime Pattern Analysis Using the Geographical

Smart Crime Pattern Analysis
Using the Geographical Analysis
Machine
Ian Turton,Stan Openshaw, James Macgill
CCG, University of Leeds
email: [email protected]
1
Crime Pattern Analysis
•
•
•
•
Automated
Smart
Easy to use
Easy to understand
2
Being SMART is not just
a matter of methodology
but also involves access,
usability, relevancy, and
result communication
factors
Residential Crimes
4
Street Crime Locations
5
6
7
Spot any patterns?
Mapping the raw data is virtually useless
unless the patterns are
blindingly
obvious
8
GAM & GEM
9
GAM creates a density
surface of weighted
evidence of clustering
which is used to suggest
locations, intensities, and
patterns of clustering that
exists on the map
10
11
12
13
GAM Results Surface
14
GAM results for Street Crime
15
GAM results for Street Crime II
16
That could be random chance!
• Each run examines 433,714 different circles
• So you might expect some circles by
random chance
• GAM lets you test that
17
Random results
18
If you want to try out WWW-GAM
http://www.ccg.leeds.ac.uk/smart/intro.html
19
But why not build
the search for local
association into the
circle search used
in GAM?
20
Building a Geographical
Explanations Machine- GEM/1
• Explanation here is to be interpreted in
the traditional geographical sense of
there being a possibly interesting
localized spatial association between
clusters and certain GIS data layers
• Maps do not cause patterns to appear BUT
they do contain clues as to the processes
that do if only we were clever enough to
spot and decode them
21
Rock D
Rock A
Rock B
Rock C
22
Geology Map
railway
2 km
23
buffer polygon
Combined Geology and Railway Buffer Map
Rock A
Rock B
2 km
Rock D
Rock C
24
Combinations of Attributes
• If we have 8 attributes with 10 classes each
• There are 3160 permutations of 2 classes
from 80 compared with 24,040,016 if any 5
are used
• Smart searches are essential
– use GA to generate possible combinations of
interest
25
Back to Baltimore
•
•
•
•
•
Visit the US Census Bureau Web site
Download Census variables at block level
Aggregate to block groups
Split variables to quartiles
Export as text files from arcview
26
House Value
27
Ethnicity
28
Old People
29
Run GEM
•
•
•
•
Similar web interface
simple ASCII text files
same visual output
I have used chloropleth maps as psuedo
coverages
• you could use other information
– distance to main roads
– neighbourhood watch areas
30
Residential Crime (Mode 1)
31
Residential Crime (mode 3)
32
Residential Crimes
• The most common combination of
coverages for clusters of residential crime
• high house values
• lots of old people
33
Street Crime
34
Street Crime II
35
Related Coverages
• For both base populations the most
commonly related coverages are
• high house values
• high proportion of white residents
36
If you want to try out Smart
Analysis on the Web
http://www.ccg.leeds.ac.uk/smart/intro.html
37
Future developments
• GAM and GEM fail eventually as more
coverages and time periods are added
• The CCG is currently developing new
methods of driving the search process
– Genetic Algorithms
– Swarm based optimization
38
Further Info: Email
[email protected]
[email protected]
[email protected]
http://www.ccg.leeds.ac.uk/smart/intro.html