14.15J | Spring 2018 | Undergraduate

Networks

Project Ideas

Project Ideas

Below is a list of project suggestions. These are merely examples and you are not required to do project from one of these topics. Use them as a “model” to potentially decide on your own project idea. Of course, you are welcome to utilize one of these topics as a project.

  1. Estimate a network optimization problem in an airline route network.
  2. Look at diffusion of ideas in patent citation data.
  3. Test homophily in a social network; look at different types of friendships and analyse interactions between them.
  4. Read two of the papers below on financial networks and contagion, and compare the approaches taken:
  5. Analyze Degroot learning by focusing on its differences from Bayesian Learning. This paper might be a good place to start:
  6. Analyze what export networks tell us about how countries grow and develop. Is the complexity measure reasonable? Is there a better measure of complexity you can come up with? As an extra empirical exercise, have a look at what import networks can tell you.
  7. Compare and contrast structural properties of different social networks. What are their common features? How do they differ?
  8. What is the structure of languages?
  9. Macroeconomics shocks: how do they spread? Review and contrast these papers:
  10. Simulate different types of epidemics in a number of (offline) social networks. Suggest prevention strategies given a fixed budget. 
  11. Apply network centrality metrics on air traffic delay networks. Identify influencers in the network and provide recommendations on how to minimize air traffic delay.
  12. What makes online content go viral? Review and contrast these papers:
  13. Download the street level traffic flow data for London (or another major city). Use data on capacity, distances (Google Maps) and speed limits to identify delays. Calculate the price of anarchy on London roads.
  14. Consider all network datasets. What minimal set of network properties do you need to know in order to classify the type of a network? E.g. how can you tell a brain network from a Facebook network without seeing any labels? 
  15. Simulate different linear best-response game in different growing networks. When does a unique equilibrium exist? What are its properties? How does it change? Who are the key players? For a simulation of a coordination game in growing networks, see:
  16. Analyze the data from memetracker. Can you identify any particular influencers? What is the structure of the network? Are there any similarities in memes/phrases that tend to be popular?
  17. Pick a few different network metrics (centralities etc) and implement them on Apache Spark using RDDs or GraphFrames. Run these on large networks—how do different centralities differ in terms of performance, parallelizability, ease/difficulty of implementation?
  18. Analyze the US power grid network. Can you identify areas particularly vulnerable to black/brown outs? Does this network resemble any of the networks we studied?
  19. Build a meeting / scheduling system (like Doodle) that accepts preference for different options in terms of comparisons from people.
  20. Use / design comparison based ranking to potentially resolve controversy of NCAA football season 1990–91.
  21. Analyze and obtain insights from the following data about connections of President (then candidate) Trump:
  22. Understand, utilize and explain “Granger Causality’.”

Course Info

Learning Resource Types
Problem Sets with Solutions
Exams with Solutions
Lecture Notes