New Viz: The Law & Order Universe

I’ve always loved the Law & Order shows. My favorite is Special Victims Unit because it has so many great characters—Olivia Benson, Elliot Stabler, and who doesn’t love Ice T as Fin Tutuola! I recently started re-watching SVU and saw the episode Design (episode 2 of season 7). The episode guest starred Estella Warren as April Troost, a victim who turned out to be a brilliant con artist, working with her mother, Lorraine Dillon (played by Lynda Carter). After conning everyone including Detectives Olivia Benson and Elliot Stabler, the final scene has Elliot telling April and Lorraine that they’ll get them eventually. With a threat like that, I knew these two guest characters would have to return at some point in the future. So, I looked them up on the Law & Order Fandom site and, sure enough, April Troost and Lorraine Dillon did make another appearance. This time, however, they appeared on the original Law & Order in an episode called Flaw (episode 2 of season 16). The show also included crossovers from Olivia Benson and Fin Tutuola. This got me thinking about the Law & Order universe as a whole. I wondered about the frequency of these types of crossovers so I decided to visualize the shows and characters.


Collecting the data was definitely tricky. The only definitive source I could find for all the shows was the Law & Order Fandom site. Unfortunately, there is no true database behind the site so I was forced to build a Python-based web scraper to scrape text from each individual episode page (nearly 1,200 individual pages). In the end, however, I had a data set that included every episode of the six US-based Law & Order shows, along with 33,000+ character references. Using this data and a fantastic blog from Chris Conn, How to use Gephi to create Network Visualizations for Tableau, I built a network diagram showing every character who appears at least twice in the Law & Order Universe. The hamburger menu in the upper right hand-corner contains some additional information about the viz as well as the ability to filter and highlight. Hope you enjoy it!



A Few Notes on the Data

The data is complete as of May 5, 2021. As noted above, I had to scrape text from each individual episode page. Unfortunately, there are some inconsistencies between pages, which led to some data cleanliness issues. Additionally, as there are no unique identifiers for characters, the name was used to distinguish discrete characters. Because some names are somewhat generic in nature, there are some instances where separate characters may be shown as the same character since they share a name. I did my best to clean up as many of these problems as possible, but with over 33,000 characters scraped, it is likely that this visualization will contain some minor errors. That said, I feel confident in the cleanliness of the major characters in the shows. If you see anything that does not seem quite right, please feel free to reach out to me.


Ken Flerlage, May 8, 2021

Twitter | LinkedIn | GitHub | Tableau Public

No comments:

Powered by Blogger.