- General Information
This is the official website of Our Netflix Project. The official webpage of the contest can be found at www.netflixprize.com. Our task is to predict future user rating on movies on existed data. It seems that some group will definitely grab the 1million dollar prize (unlikely to be us-_-), but we'll try our best for sure.
The motivation to start this task is the course project of Machine Learning Course, Fall 2007, Carnegie Mellon University. Xin Zhang(unwornmoon) and I plan to use the NetFlix data to implement some machine learning algorithms.
Our task starts at mid October, 2007. And now the best team can achieve a 0.8709 score, very close to the official prize line, 0.8563
Our course project final report can be accessed here
- Group member
Bin Fu berniefu@gmail.com Carnegie Mellon University
Xin Zhang z-x02@hotmail.com Carnegie Mellon University
Guang Xiang xguang80@hotmail.com Carnegie Mellon University
Yang Zhao zy.dennis@gmail.com Tsinghua University
Liu Liu withyliu@gmail.com Tsinghua University
- Current Progress and Ideas
Discover clusters of similar movies or users. Because the data given to us is simply users' rating to movies, so we have to use machine learning algorithms ourselves (like K-means clustering or KNN) to judge the similarity of different movies and users.
Predict the rating a user will give on a movie. Based on the similarity of users and movies, we plan to develop algorithms (like Bayesian) to predict future rating of a specific user on one movie.
Yang Zhao (Denniszz) mentioned that now the prevailing tools to do this kind of things is Collaborative Filtering (see the last paper below). Maybe we can learn something from that.
- Papers to read
James Bennett, Stan Lanning: The Netflix Prize. Proceedings of KDD Cup and Workshop, 2007.
Ted Hong, Dimitris Tsamis: Use of KNN for the Netflix Prize.
Yew Jin Lim, Yee Whye The: Variational Bayesian Approach to Movie Rating Prediction.