Here is a fun interactive display by Noah Velltman of the height and weight of National Football League players from 1920 to 2014, via Statistical Modeling... Clicking on this link allows you to vary the year and observe the joint distribution of height and weight over the years. You can even see how the game as changed from the early 1990s as roles and players have become more specialized resulting in distinct clusters of body types.
Showing posts with label clustering. Show all posts
Showing posts with label clustering. Show all posts
Monday, July 28, 2014
Monday, June 17, 2013
Using Metadata to find Colonial Traitors to the English Crown (c. 1770s)
Last week Duke sociologist Kieran Healy had a fun and informative post showing how much information about social networks can be gleaned from so little metadata. Metadata have been in the news lately with the disclosure of Prism the US Government's communication surveillance system. Healy's post shows how knowing only what groups individuals belong to and a little matrix arithmetic that individuals can be "linked through the groups they belong to [and] groups can be linked by the people they share". Using data from the book "Paul Revere's Ride" by Fischer, Healy produced the basic image above showing the links between various Colonials through the clubs, committees, caucuses, etc that they belonged to. In particular this plot shows people that serve as bridges between the groups. I've added the portrait of Paul Revere, a central bridge, "ready to ride and spread the alarm", to many political groups and marking him as a notable Colonial traitor to the British Crown of the 1770s. Healy also has a plot showing the links between the groups. So much information from so little metadata. We've used other similarity plotting techniques here to cluster the Justices of the Supreme Court.
(The follow-up to Gym Freqs, promised last week, will appear next week).
Subscribe to:
Posts (Atom)


