jessicawang225 / netflix-challenge

Jess's Angels

CS 156b

Team Members

Jessica, Bhairav, Connor, Marcus

Papers and Articles

https://www.netflixprize.com/assets/GrandPrize2009_BPC_BellKor.pdf --overview of winning solution
https://www.netflixprize.com/assets/GrandPrize2009_BPC_BigChaos.pdf --comprehensive discussion of each model in winning blend
https://pythonhosted.org/scikit-fuzzy/auto_examples/plot_cmeans.html -- Fuzzy K-means clustering

Packages

http://www.shogun-toolbox.org/examples/latest/examples/multiclass_classifier/knn.html (kNN clustering, C++ with Python hooks)
http://www.mlpack.org/docs/mlpack-3.0.0/python/cf.html (Collaborative Filtering, C++ with Python hooks)
https://cran.r-project.org/web/packages/SuperLearner/vignettes/Guide-to-SuperLearner.html (SuperLearner in R, check to download appropriate packages)

TimeSVD++

citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.379.1951

Data processing tasks

For every rating compute and save Bin(t) (all datasets)
For each user, compile a list of the unique days on which they rated (training only)
For each user, compute the mean rating date (training only)

Notes

30 time bins for function Bin(t), used in changing movie bias over time
Train using SGD for 30 epochs
Training time per epoch is roughly double that of SVD++

High predictive power
Drift in user bias over time
User per day bias (per user, this models spikes in ratings on a specific day)
Time dependence of user factors

Medium predictive power
Drift in movie bias over time

Low predictive power
Seasonal periodicity
Day of the week periodicity