swarm-modeling
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Swarm-Modelling] Agent-based model of a professional association


From: T Wallace
Subject: [Swarm-Modelling] Agent-based model of a professional association
Date: Mon, 20 Aug 2007 10:26:55 -0700 (PDT)

I've worked in the IT departments of professional
associations for many years, and when I found out
about agent-based models for simulating complex
systems, it seemed like the perfect tool for trying to
figure out what goes on at associations.

I started with NetLogo and after a few tries, ended up
with a system that showed some of the same up-and-down
cycles that I see at associations. 

Here's what I know about associations after working
for them behind-the-scenes for years:

1. If you stick with the basic three departments:
conventions, membership, and publications, you can set
up an association for just about any interest or
professional group. Take a look at all the
associations in Washington DC, for instance.

2. I learned from the marketing people I work with
that it takes seven to nine "presentations" of a
product before you can get someone to decide to buy
it. These presentations can be advertisements in your
journals, postcards, mentions in your e-newsletters,
etc.

So, turning to a more familiar tool (to me) for making
computer programs, I used VB.NET in Visual Studio 2005
to make an agent-based model of an association. Here
are some screen shots:

http://www.vindustries.com/insideassociation.html

Right now, each agent has a variable associated with
it that keeps track of the recency or effectiveness of
the last "presentation" of the association. Each agent
loses "interest" at slightly different rates. A few
agents are "true believers" who don't lose interest no
matter how few times they get a "presentation" of the
association. There is also an "impulse buy" function
that happens very infrequently immediately after a
presentation, even if the agent's total charge of
recency is low.

Right now, the boost or fade of the variable that
tracks recency-of-presentation is linear, but I
suspect that there are better representations of
people's response to advertising out there -- maybe
the sigma function we use in neural networks? Any
others?

Thanks for your comments and interest,
Todd Wallace


      
____________________________________________________________________________________
Park yourself in front of a world of choices in alternative vehicles. Visit the 
Yahoo! Auto Green Center.
http://autos.yahoo.com/green_center/ 


reply via email to

[Prev in Thread] Current Thread [Next in Thread]