Written by Scott D. Anderson, © 2004
Scott.Anderson@ACM.org
Creative Commons License
This work is licensed under a Creative Commons License.

Project Goals

The primary goal for the project is to create a simulation model of some system of your choosing. It could be from economics, international relations, physics, you name it. It needn't be from your major, or even from any academic subject. What it needs is:

  • To be quantitative. Somehow we must put numbers on something. You can use either Extend, as most people will, or StarLogo, if you'd like to do some programming.
  • To be interesting and challenging. This project counts like two problem sets and I'm allowing more than two weeks. Your project should be something appropriate for that time frame.

The model doesn't necessarily have to be very complex. It's possible that it can be a fairly simple model where instead you've put your effort into researching the constants, equations and probability distributions that underly it. For example, it would be cool to model, say, Schneider, and actually count arrival rates and service times.

As secondary goal is to statistically analyze data from your simulation model. This might be no more than a point estimate of some quantity, along with a confidence interval. Or, you might have a simulation that can compare two ways of doing something, and you can do a t-test to decide which is better.

Phase 1

At the beginning, I want an email message describing your project. I would be very glad to meet with you to discuss the project in more detail. I will typically include this along with one of the last few homeworks. In that description, you'll give the jist of your idea, whether the model will be continuous or discrete event, the factors that go into the model, what is random versus constant, and so forth.

Phase 2

At about the midway point, I'm requesting an email message that briefly summarizes how your project is going. You can send a rough draft of the simulation model if you have one, you can tell me what equations and distributions you've learned about in your research, and so forth. Try to determine whether you're on track or have fallen behind, and how I can help (if I can). This is more for your benefit than mine, just to give you a chance to assess your own progress and determine a plan for finishing in time.

Of course, it can't hurt to use this as an opportunity for a rough draft, so certainly tell me as much as you can.

Final Submission

Please submit an email to the drop folder with your model attached. Either your email or your model (preferably the latter) should tell me the background of the model, the assumptions underlying it, the equations you're using, citations for sources, and so forth. By reading this documentation, I should be able to understand your model. Also, include any directions for running the model, if necessary.

You should put a little effort into laying out your model in a neat, clear manner. Colored areas and such can help in understanding the structure of your model. There's no need to go overboard with this or to spend a lot of time making it "pretty," but aesthetics can sometimes be functional, and this is one such case.