It turns out I’m probably going to have good reasons to need to learn Python in the near future, so I’m finding little projects to practice writing scripts. The Incomparable recently wrapped up the year with a retrospective episode in which some people talked about being on more than in the past, while others mentioned they were on less. I was inspired to see if there was a way to visualize who has been on episodes together and how that has changed over time.
I collected data from the RSS feed and the web site to figure out which people were on which episodes. I could then put that into a matrix with episodes as the rows and people as the columns (with 1’s or 0’s depending on whether they were on the episode or not). Multiplying the transpose of that matrix by itself gives a new matrix that shows how many episodes people had in common. (This is just like what Kieran Healy did in his Paul Revere metadata post.)
I could then use D3js to visualize the results as a force-directed graph. I calculated the network matrices for subsets of episodes from different timeframes, so you can flip through and see how the graph changes over time. You can see the results here.