Neat! I'm attempting to generate a ranking for Seattle now. [Edit: Nevermind, this Java newbie got bored trying to figure out jars and classpaths.] I also didn't realize before that you could find users by their most-used language:
I meant that I was attempting to install and run the code from the article, which is written in JRuby+Java and depends on a couple of third-party Java libraries.
It's way easier to compute betweenness centrality? Eigenvector centrality requires inverting a potentially giant link matrix -- a notoriously expensive operation, which says nothing of the additional code complexity needed to do it properly.
You can get the first eigenvector through power iteration, just take some random vector then repeatedly normalize it and multiply by the adjacency matrix. So no giant matrix inversion needed.
14 comments
[ 5.2 ms ] story [ 51.8 ms ] threadhttp://github.com/search?type=Users&language=python&...
...and GitHub's own results are ranked by total number of followers, which gives a simpler version of this article's "connectedness" measurement.
The full lists for SF and London are here: http://github.com/mattb/flotsam/tree/master/github-recruitme...
Looks like I'm languishing at #27 in the London list, better get back to work..