Because I just couldn’t get enough of the new Machine Learning Specialization from the University of Washington, I decided to fill fill my schedule to the brim with another Coursera class, Social and Economic Networks: Models and Analysis, from the University of Stanford. I took a graph theory course at the University of Illinois while getting my master’s degree around the dawn of the new millennium, which among many other topics, covered things like Euler circuits, Hamiltonian paths, coloring, and the like.
I’ve had a few occasions to revisit that material when teaching graph theory in discrete math courses with high school students and participating in mathematics research with undergraduates at Illinois State University. However, I’ve had a sense that the important algorithms, metrics, and even basic vocabulary have shifted as things like social media networks have emerged, and it was time to start getting caught up.
For true beginners, the vocabulary of graph theory is covered in week one before diving into the more sophisticated concepts that comprise the rest of the course. Weeks two and three focus on formation of random graphs, with an emphasis on Eros-Renyi Random Graphs. As he advances through the course, Stanford professor Matthew Jackson explains graph models including strategic network formation, diffusion on networks, and ultimately learning on networks (full syllabus here).
Jackson does an excellent job relaying a significant amount of material over the course of this 7 week class. His explanations are understandable, and while he delves into a fair amount of theory, he is also good about giving concrete small scale examples of the graphs to illustrate the application of the concepts he is explaining. Jackson also offers deeper dives into proofs of concepts at various points throughout the course but makes those videos optional to completing the assignments if you don’t care to go quite so deep.
Throughout the course, Jackson stuck to the use of Gephi. I had used Gephi a bit before but enjoyed getting to learn more about how it works here. It is a great tool that is easy to use and visualize graphs with. It would have been even better to see some of the lessons incorporate NetworkX with Python or iGraph with R to see how some of these concepts are implemented with these very popular languages. However, I do realize that by sticking to a point and click tool like Gephi that the class remains accessible to a wider audience.
Bottom Line Review
The only real negative I can find with this class is that the discussion board was not very active. I never saw enrollment numbers released for this class, but I can only assume that with few people discussing the class, there must not have been more than a few thousand people in the class. That is a shame because Matthew Jackson has done a great job creating a course that is very relevant in today’s world, and I hope more people will catch it when he offers it again. He mentioned in a discussion thread that he plans to do so sometime in the Spring of 2016.
- Regression Models Coursera Review
- Coursera Review: Machine Learning Foundations—A Case Study Approach
- Getting and Cleaning Data Coursera Review