Re: [Swarm-Modelling] Modelling robbery through Agent based modelling

[Steve Railsback]

Paulsen, Derek wrote:
> Swarm Listserve,
>      I am a professor of Criminal Justice who has done quite a bit of 
> research on spatial patterns of crime, particularly commercial robbery 
> (robbery of commercial locations) and I am wanting to try and create a 
> simulation of commercial robbery for an actual city.  My attempt is to 
> see if I can model/simulate robbery in a real city to test the impact of 
> different factors (location, security, police presence) on robbery 
> offenses.  I have done a considerable amount of research on spatial 
> aspects that impact victimization as well as offender target selection 
> characteristics (interviewed convicted offenders) and thus I have a good 
> theoretical model for how commercial robbery locations are selected by 
> offenders.  My question(s) is where is a good place to start?  I have 
> very little programming experience, but have access to those who do.  Is 
> there any particular system that integrates better with GIS and might 
> work better for me?  What is the best way for me to go about learning 
> more about the creation of models?  Is this a social process that can be 
> simulated/modelled or am I off base? Any assistance would be greatly 
> appreciated. 

Hi- It's good to see new people posting messages like this.

Your idea sounds like the kind of model that might work well as an 
agent-based model not because it must be mathematically (e.g., because 
interactions and variability among agents are important) but because it 
is the easiest way to describe and understand the problem.

There are a number of groups working on agent- and GIS-based modeling of 
social problems. The Center for Study of Complex Systems at U. Michigan 
is modeling urban spawl, for example. And the Repast program started as 
an agent-based modeling support group in the Social Sciences department 
at Univ. Chicago. If you are not familiar with Repast 
(http://repast.sourceforge.net/), it is a Swarm-like agent-based 
modeling platform now housed at Argonne National Lab.  I believe they 
have incorporated some GIS capabilities but I am not yet familiar with 
them.

How to get started...I always encourage people to think quite a bit 
about their model and its design (what problems is it intended to solve? 
what information is available to build and test it? what processes and 
variables and behaviors absolutely must be included?) before worrying 
about programming.

I also think it is often a good idea to start with an extremely simple 
model implemented in NetLogo, an extremely simple modeling platform 
(that comes with a collection of models, some of social problems, as 
examples).

I don't know what modeling literature there might be in your field. My 
colleague Volker Grimm and I have a new book on individual-based 
modeling, that starts with several chapters on the modeling process in 
general and how to get started. Our book is focused on ecology but 
should be useful to people in other fields. It will be available in a 
few weeks; see: http://www.humboldt.edu/~ecomodel/book.htm

Steve Railsback
--
Lang Railsback & Assoc.
250 California Ave.
Arcata, California 95521
707 822 0453