Agent Based Modelling Example

Understanding and preventing burglary

Dr Nick Malleson

Dr Andy Evans

[Fullscreen]

Theoretical Backgound

Shift in (environmental) criminology towards 'opportunity theories'

Focus on social / environmental factors surrounding an individual crime

Major prevalent theories:

Routine activities theory

Geometric Theory of Crime (Crime Pattern Theory)

Rational Choice Perspective

Theoretical Backgound

Routine Activities Theory

Routine activities theory

Three elements converge:

A motivated offender

A victim

The absence of a capable guardians

Explains increasing crime rates in 60s/70s (?)

Cohen and Felson (1979)

Theoretical Backgound

Geometric Theory of Crime

Routine activities theory

Everyone has a cognitive map (‘awareness space’) of their environment

This is built up from travel around ‘anchor points’.

People will commit crimes in areas they know well and feel safe in – their awareness space

Brantingham & Brantingham (1981)

Theoretical Backgound

The Rational Choice Perspective

Routine activities theory

A framework for understanding criminal choice

(Bounded) rationality

Cost-benefit analysis

Clarke and Cornish (1985)

Theoretical Backgound

Example of regression

Modelling Crime

Theories are about individuals

Complex micro-level interactions of individuals and environment

E.g. burglary:

Individual houses: visibility of properties, burglar alarm, back door etc

Individual burglars: feel "safe" in neighbourhood?, aware of opportunity?, drug addiction?, "professional" or "opportunist"?

Problems with aggregate models

A Solution: Agent-Based Crime Modelling

The Sims

Rather than controlling from the top, try to represent the individuals

"Grow" the phenomena from the ground up (Epstein and Axtell, 1996)

'Object-based models' - variables and equations are "encapsulated" within objects (Cioffi-Revilla, 2014)

Account for system behaviour directly

Can model emergence, non-linearity, and other features of complex systems

Appeal of ABM

Most 'natural' way of thinking about social systems

Physical space / social processes

Designed at abstract level: easy to change scale

Bridge between verbal theories and mathematical models

Dynamic history of system

Disadvantages of ABM

Known unknowns

We don’t know exactly what someone will do, so we develop probabilistic models

E.g. Jacob's choice: Duck or Truck?

Computationally expensive

Complicated agent decisions

Lots of decisions!

Multiple model runs (robustness)

Modelling "soft" human factors

A benefit is that we can include complex psychology

But this is really hard!

DATA!

(more on this later)

Burglary Simulation

Model Overview

Aim: Build a spatially-realistic model of burglary movements in Leeds

Able to explore the impacts that changes to the environment or changes in behaviour have on crime in a real area

Two main components:

Virtual burglars (agents)

The virtual environment

Burglary Simulation

Virtual Environment

Communities, Buildings, Roads

Mastermap topography

Burglary Simulation

Burglar Behaviour: PECS

Physical Conditions, Emotional States, Cognitive Capabilities, Social Status

State variables, intensity functions, motives, action-guiding motive

PECS
The burglars' decision process

Burglary Simulation

Burglary Decision Process

decision process

www.youtube.com/embed/5ySbS075MyA

Model Calibration

Compared the model to real burglary data

Manually adjusted rules to match known data (calibrate)

Future work: automatic calibration (e.g. with artificial intelligence algorithms)

comparing to real burglary data

An interesting finding...

model errors in Halton Moor

Model Experiments

Urban regeneration - EASEL

Very large scheme, involving changes to:

Houses

Communities

Roads

Aim: expore the impacts that the changes might have on residential burglary in the area

Experiments based on real plans

Results of the burglary simulation

Conclusion

new york at night

Theoretical focus on the individual

ABM captures these individual-level dynamics

'What-if' simulation with ABM

Future work:

Heterogeneous behaviour

Better validation with pattern oriented modelling (need data!)

Non-offender behaviour (need data!)