Sunday, June 01, 2008

Right Data vs. Better Models

Yehuda Koren, a member of the famous BellKor team in the Netflix Prize competition, recently wrote an article about the three levels of addressing the Netflix Prize. The insight that deciding what data to model is more important than choosing what model to use seems to be complement to the issue of more data vs. better algorithms. This is also supported by the successful story of Amazon's recommendation systems.

No comments: