Congressional Network Analysis by Christopher Roach.
From the post:
This page started out as a bit of code that I wrote for my network science talk at PyData 2013 (Silicon Valley). It was meant to serve as a simple example of how to apply some social network analysis techniques to a real world dataset. After the talk, I decided to get the code cleaned up a bit so that I could release it for anyone who had seen the talk, or just for anyone who happens to have a general interest in the topic. As I worked at cleaning the code up, I started adding a few little features here and there and started to think about how I could make the visualization easier to execute since Matplotlib can sometimes be a bit burdensome to install. The solution was to display the visualization in the browser. This way it could be viewed without needing to install a bunch of third-party Python libraries.
Quick Overview
The script that I created for the talk shows a social network of one of the houses for a specific session of Congress. The network is created by linking each member of Congress to other members with which they have worked on a at least one bill. The more bills the two members have worked on, the more intense the link is between the two in the visualization. In the browser-based visualization, you can change the size of the nodes relative to some network measure, by selecting the desired measure from the dropdown in the upper right corner of the visualization. Finally, unlike the script, the graph above only shows one network for the Senate of the 112th Congress. I chose this session specifically simply because it can be considered the most dysfunctional session of congress in our nation’s history and so I thought it might be an interesting session for us to study.
I think the title for “most dysfunctional session” of congress is up for grabs, again. 😉
But this is a great introduction to visualization with D3.js, along with appropriate warnings to not take the data at face value. Yes, the graph may seem to indicate a number of things but it is just a view of a snippet of data.
Christopher should get high marks for advocating skepticism in data analysis.