Re:Nate Silver: What I need from statisticians
by micro_cam on Tuesday October 29, 2013 @12:02AM
(#6631942)
It isn't unusual for a data analysis algorithms to not be shown to work theoretically until after it is published. Ie here is the first proof that adaboost converges some 3 years after the paper that introduced it:
http://l2r.cs.uiuc.edu/~danr/Teaching/CS598-05/Papers/Collin...