Transcript SDM-Crime Analysis - University of Minnesota Twin Cities

Spatial Frequent Pattern Mining
for Crime Analysis
Application Questions

Crime analysis
Localizing frequent crime patterns, Opportunities for crime
vary across space!
Question: Do downtown bars often lead to assaults more
frequently ?

• Law enforcement planning
Question: Where are the frequent crime
routes ?
Courtsey: www.startribune.com
Forecasting crime
levels in different
neighborhoods.
• Predictive policing (e.g. forecast crime levels
in different neighborhoods )
Question: What are the crime levels 1 hour after
a football game within a radius of 1 mile ?
2
Scientific Domain: Environmental Criminology
Crime pattern theory
Routine activity theory
and Crime Triangle
Courtsey:
http://www.popcenter.org/learning/60steps/index.
cfm?stepnum=8
Courtsey: http://www.popcenter.org/learning/60steps/index.cfm?stepNum=16
Courtsey: www.amazon.com
 Crime Event: Motivated offender, vulnerable victim (available at an appropriate location
and time), absence of a capable guardian.
 Crime Generators : offenders and targets come together in time place, large gatherings
(e.g. Bars, Football games)
 Crime Attractors : places offering many criminal opportunities and offenders may
relocate to these areas (e.g. drug areas)
6
Spatial Frequent Pattern Mining
Process of discovering interesting, useful and non-trivial patterns from spatial data.
4
Illustrative Frequent Patterns: Regional Co-location
 Input: Spatial Features, Crime Reports.
 Output: RCP (e.g. < (Bar, Assaults), Downtown >)
 Subsets of spatial features / Crime Types.
 Frequently located in certain regions of a study area.
Larceny, Bars and Assaults
Q. Are downtown Bars likely to be more
crime prone than others ?
Dataset: Lincoln, NE, Crime data (Winter
‘07), Neighborhood Size = 0.25 miles,
Prevalence Threshold = 0.07
N
Observation : Bars in Downtown
are more likely to be crime prone
than bars in other areas (e.g.
20.1 % Shown by blue polygon
area).
5
Illustrative Frequent Patterns: K Main Routes
 Input: Crime Reports, Road Network, K (# of Patrol Vehicles)
 Output: K- Main Routes Taken by the Patrol Vehicles
Dataset: U.S. City (Southern U.S), K = 10
N
K- Main Routes
K- Main Routes / CrimeStat ellipses
6
Illustrative Frequent Patterns: Crime Outbreaks
 Input: Crime Reports, Crime Types, Spatial Features (Bars)
 Output: (a) Bars with more than usual crime activity, (b) Crime Types that are highly
active around bars, (c) Regions (Crime Outbreaks) around Bars with high risk of crime.
N
Vandalism Crime Outbreaks around Bars.
Alcohol crime outbreaks around bars.
Legend: (a) Risk Region Represented by Red Circle; (b) Black stars (*) represent Bars7
Crime Outbreaks to Regional Crime Patterns
 Input: Crime types involved in a large number of significant Crime Outbreaks
(Slide 7’s output)
 Output: Regional co-location patterns between crime types involved in one or
more outbreaks.
Dataset: Lincoln, NE, Crime data (2007),
Neighborhood Size = 700 feet, Prevalence
Threshold = 0.001
Observation : Bars in Downtown
have a marginally higher chance
(4.6%) to witness Alcohol as well
as Vandalism related Crime
Outbreaks (Center Polygon).
9
Spatio-temporal Frequent patterns: Cascading Patterns
N
Lincoln, NE crime dataset: Case study
 Is bar closing a generator for crime related CSTP ?
N
Bar locations in Lincoln, NE
Questions
 Does Crime Peak around bar closing ?
Observation: Crime peaks around bar-closing!
Bar closing
Saturday Night
Increase(Larceny,vandalism, assaults)
Increase(Larceny,vandalism, assaults)
References
 S. Shekar, P.Mohan, D.Oliver, X. Zhou. Crime Pattern Analysis: A spatial frequent pattern mining
approach. Department of Computer Science and Engineering, University of Minnesota, Twin-Cities,
Tech Report (TR 12-015), URL: http://www.cs.umn.edu/tech_reports_upload/tr2012/12-015.pdf
 P.Mohan, S.Shekhar, J.A. Shine, J.P. Rogers, Z.Jiang, N. Wayant. A spatial neighborhood graph
approach to Regional Colocation Pattern Discovery.
 D. Oliver, A. Bannur, J.M. Kang, S.Shekhar, R. Bousselaire. A K-Main Routes Approach to Spatial
Network Activity Summarization. ICDM Workshops 2010: 265-272
 P. Mohan, S. Shekhar, J.A. Shine, J.P. Rogers. Cascading Spatio-temporal Pattern Discovery. In
IEEE Transactions on Knowledge and Data Engineering, 2012, November (to Appear).
 Jung I,Kulldorff M,Richard OJ,. A spatial scan statistic for multinomial data . Stat Med. 2010 Aug
15;18:1910-1918
12