Guidelines for the final project
The final project is a self-driven agent-based modeling analysis that culminates in a paper and oral presentation.
The final paper and your model code will be due the day of the final exam.
Project guidelines
- You may work alone or in pairs (three can work together but talk to me first since it can be hard to parse how much effort everyone put in)
- This project should be original work, but you are encouraged to tie in your own research or
research interests—if this project can eventually lead to a paper or dissertation chapter that’s great! (Just be sure that it’s also something do-able in one semester!)
- I will periodically check in with you/your group throughout the semester to make sure you’re on track and that your project has a reasonable scope. There will be benchmarks included as part of some of
the labs as well.
Paper
For the final paper, you will need to provide:
- The write-up: I anticipate this will likely be 8-12 pages, but shorter or longer is fine so long as you fully cover the pieces described below
- Your model: I should be able to both review the code and run it myself if in NetLogo or Python. Your code must be documented, clear, and readable. Be sure to also document the version of python and versions of any packages you used.
Paper Components
- Introduction. Provide an overview of the problem and a literature review. You should address: what gap or question are you addressing? What has been done before?
- Methods
- Model description. Describe how your model works in terms of its: agents, interactions, environment, model schedule/timing
- It’s a good idea to use the PARTE and ODD frameworks as a guide
- Flow charts & visuals are good! Illustrate the sequence of events in your model using a flow chart. Show how the agents operate, how and when interactions happen, the sequence of events in each time step, etc.
- From your model description, someone should be able to implement a version of your model
- Model analysis. Describe the parameter settings you swept through and the analyses you ran. Someone should be able to recreate your analyses from your description, so be specific and complete!
- If your model is stochastic, you may need to run multiple trials at the same parameter settings.
- Results. Provide qualitative and quantitative summaries of how your model behaves. Provide graphs and plots of model outcomes at different settings as needed.
- Discussion. Return to the question or problem in your introduction—what do your results say about this problem? Put your results in a broader context. Describe the strengths, limitations, and potential future directions of your work.
- If you think there are still some bugs driving your model’s behavior, this is the place to discuss this
- Also a good place to talk about how you might verify, validate, or extend the model in the future