Synthesis over Analysis: Using Multi-Agent Simulations to Examine the Interactions of Crime Dan Birks Justice Griffith Griffith University.

Slides:



Advertisements
Similar presentations
SETTINGS AS COMPLEX ADAPTIVE SYSTEMS AN INTRODUCTION TO COMPLEXITY SCIENCE FOR HEALTH PROMOTION PROFESSIONALS Nastaran Keshavarz Mohammadi Don Nutbeam,
Advertisements

Psychlotron.org.uk You have two minutes to draw a map of Birmingham.
MODEL ACADEMIC CURRICULUM MODULE 6 Situational Crime Prevention
NTU Criminology Dr Jason Pandya-Wood.  Crime Reduction and Policing: investigates our understanding of crime and the effectiveness of policy responses.
TERRITORIAL FUNCTIONING AND VICTIMISATION IN COUNCIL ESTATES IN SHEFFIELD By: Aldrin Abdullah.
Chapter Four The Development of Rational Choice Theory
Choice Theory Development of Choice Theory Concepts of Rational Choice
Using NIBRS (and other data) for Understanding the role of offender’s criminal history in violence against the police Donald Faggiani, University of Wisconsin.
Exploring place and time dynamics of vandalism in a Scottish Context.
Understanding and preventing crime: A new generation of simulation models Nick Malleson and Andy Evans.
The Sociological Point of View Examining Social Life
Yiannis Demiris and Anthony Dearden By James Gilbert.
1 MODEL ACADEMIC CURRICULUM MODULE 5 Crime Theories and Crime Opportunity.
Perpetuity Research & Consultancy International (PRCI) Ltd Is CCTV working? The effectiveness of CCTV and the British experience.
Justice Griffith Family Youth Conferences and Indigenous Over-representation: Micro Simulation Case Study Anna Stewart.
Problem Identification
Simulation Models as a Research Method Professor Alexander Settles.
Police Technology Chapter Twelve
Research problem, Purpose, question
Lars-Erik Cederman and Luc Girardin Center for Comparative and International Studies (CIS) Swiss Federal Institute of Technology Zurich (ETH)
RESEARCH DESIGN.
Rational Choice, Routine Activities, and Deterrence.
© 2001 Vito & Blankenship. Learning Objectives In this chapter you will learn role of statistical analysis in criminal justice how crime in measured in.
Predicting Burglary Hotspots Shane Johnson, Kate Bowers, Ken Pease.
“Rational Choice” and Opportunity Theories. “Rational Choice Theory” Economics (language, theory) –“Expected Utility” = calculation of all risks and rewards.
Linking multi-agent simulation to experiments in economy Re-implementing John Duffy’s model of speculative learning agents.
Rationality meets the tribe: Some models of cultural group selection David Hales, The Open University Hales, D., (2010) Rationality.
Situational Crime Prevention Understanding Criminology Dan Ellingworth Tuesday, 17 th February 2009.
RATIONAL THEORIES. Overview of Rational Theories Late 1970s – 1980s Outgrowth of victimization data More conservative criminology Borrowed from economic.
Virtual Models English 197 Jennifer Yang. Reality Bytes - Casti To what degree can we employ electronic surrogates for the reality that they represent?
NAVEEN AGENT BASED SOFTWARE DEVELOPMENT. WHAT IS AN AGENT? A computer system capable of flexible, autonomous (problem-solving) action, situated in dynamic,
“Rational Choice” Theories and Situational Crime Prevention
25 Techniques of Situational Crime Prevention
Copyright © 2012, 2009, 2006, 2001, 1997 by Pearson Education, Inc. Upper Saddle River, New Jersey All rights reserved 0 Criminology: A Sociological.
1 Research methods and models of driver behavior studies.
Digital Intuition Cluster, Smart Geometry 2013, Stylianos Dritsas, Mirco Becker, David Kosdruy, Juan Subercaseaux Welcome Notes Overview 1. Perspective.
Jon Froehlich 1 Leah Findlater 2 James Landay 1 1 Computer Science and Engineering (DUB Institute, University of Washington ) 2 The Information School.
1 Real-Time Parking Information on Parking-Related Travel Cost TRIP Internship Presentation 2014 Kory Harb July 24, 2014 Advisor: Dr. Yafeng Yin Coordinator:
Developing the theoretical and conceptual framework From R.E.Khan ( J199 lecture)
12/17/ Geographic Profiling Geographic Profiling Charles L. Feer, JD, MPA Bakersfield College.
FBI Method of Profiling Violent Serial Offenders
Professor Alex Hirschfield, HonMFPH Professor of Criminology and Director International Centre for Applied Criminology ICAC University of Huddersfield,
Crime Risk Models: Specifying Boundaries and Environmental Backcloths Kate Bowers.
Forensic Victimology 2nd Edition Chapter Six: Victim Lifestyle Exposure.
Cost effectiveness in Human Services: simple complicated and complex considerations Towards a method that is useful and accurate Andrew Hawkins, ARTD Consultants.
ABRA Week 3 research design, methods… SS. Research Design and Method.
Unit 4 Dr. Marie Mele. Topics to Discuss Ability of people to make rational choices How people weigh the risks and rewards of engaging in crime How the.
Important Piece of Research I Forgot to Cover Larry Sherman’s Domestic Violence Experiments  Random assignment of police response to D.V. Counsel Separate.
Routine Activities Theory. Explaining Crime Patterns Increase in household burglary Lower crime rates in the 1990s Higher levels of crime at.
You have two minutes to draw a map of Birmingham..
Routine Activities and Rational Choice Theories. Routine Activities Theory Direct contact predatory violations Illegal activities feed off of legal activities.
Crime theory.
Situational Crime Prevention Understanding Criminology Dan Ellingworth Tuesday, 17 th February 2009.
Application of Criminological Theory to Terrorism Prevention Detective Eric Paull Planning Research and Development Akron, OH Police Department.
Introduction to Machine Learning, its potential usage in network area,
Prevention Analysis Security Services.
1st November, 2016 Transport Modelling – Developing a better understanding of Short Lived Events Marcel Pooke – Operational Modelling & Visualisation Manager.
Health Education THeories
Analytics and OR DP- summary.
Rational Choice Theory
What works? Key outcomes from a systematic inventory of 260 meta-evaluations within three knowledge domains: Situational Crime Prevention, Policing and.
Intelligent Agents Chapter 2.
Chapter Four Rational Choice Theory
Routine Activities Theory
CASE − Cognitive Agents for Social Environments
Ron Clarke- Rational choice theory
Week One - Review.
Environmental Criminology
Environmental Criminology and Crime Science
Crime Prevention Ottawa: Community Safety and Personal Safety
Presentation transcript:

Synthesis over Analysis: Using Multi-Agent Simulations to Examine the Interactions of Crime Dan Birks Justice Griffith Griffith University Justice Modelling Workshop – July 2008

Overview  What are multi-agent simulations?  Multi-agent simulations & crime analysis  A simple example – Cops & Robbers  Advanced example - MAS of Volume Crime  Some initial results  Potential Applications

What are Agent-based Simulations?  Aim to model complex systems.  Offer a “bottom-up” approach which concentrates on the study & replication of micro-level interactions which produce the macro level outputs we observe.  “Thought experiments” or “Intuition pumps” allow us to examine the ramifications of our theoretical assumptions.  Allow us to attempt to bridge the gap between theory and observed phenomena.

 Definition from artificial intelligence: Autonomous program situated within some simulation environment.  Agents perceive, reason and act  In order to do so agents have:  Internal representations (memory or state)  Method for modifying internal representations (perceptions)  Methods for modifying environment (behaviours) What is an Agent?

Agent-based Modelling & Crime Analysis  Large proportion of conventional crime analysis is “top-down” involving examination of crime or crime patterns at the macro level.  Large proportion of theory is positioned at the micro level.  A gap exists between observed macro level crime patterns and the micro level mechanisms theories hypothesise about.  ABM: tool to test the ramifications of theoretical assumptions by creating a population of virtual offenders, guardians and targets and bestowing upon them behaviours defined by our theories.  We can then examine the emergent properties of our simulation and evaluate whether our theories are causally sufficient to explain the macro level phenomena observed in the real world

TheoryFormalism Agent behaviourAgent behaviour Environment dataEnvironment dataModel Compare with real data Test Refine Simulation Methodology Theory – formalism – test - refine

A simple example: victimisation & detection Cops & Robbers  Imagine we want to examine the following theories of victimisation & detection: –A victimisation occurs when an offender comes into the same location as a potential target in the absence of a capable guardian. –A detection/prevention occurs when an offender, potential target and guardian all come together at the same point in space and time. Victimisation / (x,y,t) Suitably Motivated Offender Suitable Target Absence / Presence of Capable Guardian

Prevention/Detection Occurs if(is_present(x,y,t,offender) & is_present(x,y,t,target) & is_present(x,y,t,guardian) Prevention/Detection Occurs if(is_present(x,y,t,offender) & is_present(x,y,t,target) & is_present(x,y,t,guardian) Crime occurs if(is_present(x,y,t,offender) & is_present(x,y,t,target) & not(is_present(x,y,t,guardian)) Crime occurs if(is_present(x,y,t,offender) & is_present(x,y,t,target) & not(is_present(x,y,t,guardian)) Cops & Robbers From thought experiment to simulation A crime occurs when an offender comes into the same location as a potential target in the absence of a capable guardian. A detection/prevention occurs when an offender, potential target and guardian all come together at the same point in space and time.

Cops & Robbers – The Simulation implementation in NetLogo Person (Potential Target) Cop (Capable Guardian) Robber (Motivated Offender)

Multi Agent Testbed for Volume Crime Activity

Routine Activity Theory (Felson 1979) Crime(space,time) Absence of Capable Guardian Motivated Offender Suitable Target Rational Choice Theory (Clarke and Cornish 1985) “sees criminal behaviour not as a result of psychologically and socially determined dispositions to offend, but as the outcome of the offender’s broadly rational choices and decisions” Some background theory… Target Areas Activity Space WorkRecreationHome Offender search patterns and personal activity space Home to work to recreation – nodes and paths, and mental maps Looking for opportunities (which are non-uniformly distributed) Templates or schemas for successful offending developed Crimes in areas where offenders activity spaces overlap with target areas Crime Pattern Theory (Brantingham and Brantingham 1993)

Multi-Agent Testbed & Theories of Crime Aim: Victim - Offender - Location  To develop a multi-agent test-bed which enables the examination of Victim - Offender - Location interactions  Offender Data –Offender RAT Nodes, Propensity, etc. –Offender awareness space –Offender Behaviour (bounded rationality) Schemas for offending by type  Geographical ‘back cloth’ data –Simulate location & environment Geo-demographic Data Transport network Investigate “Victim – Offender - Location” interactions Target Areas Activity Space WorkRecreationHome

Multi-Agent Testbed for Volume Crime Activity Combining the three “Opportunity” theories we might say that When an individual of a certain criminal disposition, going about his/her routine activities, comes into the location of a suitable target, which provides an opportunity for gain and which he/she is aware of and capable of exploiting, and when he/she perceives the reward to be sufficient to expend the effort required and endure the risks involved, he/she will commit the offence at the current point in space and time.

Crime occurs when Perception of Opportunity > threshold(propensity,readiness) Where Perception of opportunity = Awareness_of_opportunity * Target Density * capability_to_exploit(Opportunity(Target))AND Offender_percieved_reward(Opportunity) > Percieved_Risk(Opportunity) * Effort(Opportunity)

Offender Agent Specification  Characteristics  Propensity / Lambda (Dynamic)  Readiness / Current desire (Dynamic)  Awareness Space (Dynamic)  Agent Behaviours  Navigation Agents dynamically navigate the street network building up awareness space. Choice of route based on simple heuristics / greedy algorithms  Identify the shortest/most direct routes  Awareness of route  The choice to offend  Awareness of location and opportunities  Perception of Opportunity/Risk/Reward

 Static targets  Relatively good geographic data  Matches well with geodemographic data –Acorn (A classification of residential neighbourhoods) –IMD (Index of multiple deprivation)  Existing observable & potentially replicable phenomena such as Repeat Victimisation and Near Repeats  Good existing metrics which allow quantitative comparison of simulation & real-world output, i.e. Knox / Mantel An example offending schema: Domestic Burglary

Offending schema: Domestic Burglary  Crime-specific prerequisites:  A domestic property  Crime-specific risk factors:  Occupancy of property; presence of deterrent measures (alarms etc); pedestrian footfall in the vicinity.  Crime-specific reward factors:  Affluence of area; likelihood of property containing „CRAVED ‟ goods.  Crime-specific effort:  Security of property in question (e.g., door locks); layout of property (e.g., back alleys).

Simulation Environment Formalism  Sufficiently detailed to allow behaviours to draw upon enough data to encapsulate theories from environmental criminology  Geo-descriptive data  Housing Density  Transport network nodes, paths  Road Capacity  Geo-demographic Data  Pedestrian Footfall  Deprivation Indicators  Household makeup

An Example Environment Residential Property Commercial Property

BUILDING OCCUPANCY – T1 BUILDING OCCUPANCY – T2 RESIDENTIAL HOUSING DENSITY How might we formalise the environment TRANSPORT NETWORK

An Example Environment

 Real Environmental Data  Road Network  Residential Address Points  Hypothetical Offender Data  X number of agents created  Agents randomly allocated 4-6 routine activity nodes  Agents randomly allocated journeys (e.g. home  work; work  leisure)  Agents randomly allocated propensity rates Simulation Demo Initial Conditions

Initial results: Emergent properties  Remember our aim: Simulate the interactions of victim-offender-location - if our simulation of behaviour is accurate, then realistic crime patterns should emerge.  Initial results show the presence of Repeat and Near-repeat victimisation. Simulation crimes produce similar profiles to that of actual residential burglary data.

Potential Applications  Academic Applications  Examine, test & refine current criminological theory  Practitioner Focused Applications  Intervention prototyping  Evolving optimal deployment of resources  Offender routine activity profiling  Educational Applications  Visualisation of theory

Relevant Publications: Birks, D.J., Donkin, S., Wellsmith, M. (2007) Synthesis over Analysis: Towards an ontology for volume crime simulation. In John Eck & Lin Liu (Eds.), Crime Analysis Systems: Using Computer Simulations and Geographic Systems. Idea Group PLC Groff,L., Birks, D.J., (2008) Simulating Crime Prevention Strategies: A Look at the Possibilities. Policing - A journal of Policy and Practice Townsley,M., Birks, D.J., (In Press) Building Better Crime Simulations: Systematic replication and the introduction of incremental complexity. Journal of Experimental Criminology “Simulated Experiments in Criminology and Criminal Justice” Birks, D.J., Eck, J., (Forthcoming) Neighbourhood Differences in Crime May Be the Result of Individual Connectivity (not all that other stuff you were taught)

Questions & Contact Dan Birks Justice Griffith Griffith University