The annual Progress Prize for the Netflix contest has been announced. The prize was won by Team BellKor, two researchers from AT&T Research who “blend[ed] 107 individual results” to come up with their 8.5% improvement over Netflix’s Cinematch algorithm (Thanks to Dan and Sujan dai for the pointers).
This contest was part of my senior project at college, where we looked at a number of algorithms to try to predict movie ratings using Netflix’s data (and use them to generate movie recommendations), and also generate recommendation of courses for students at Carleton, using historical transcript data.