Ideas

Partition
- Split users into "light" and "heavy" users, based on how many movies they've rated total.
- Compute weighted average, giving more weight to the "heavy" users
Remove users who only rate movies a 1 or 5 (extreme users)
Calculate how much higher/lower a user typically rates their movies
- for a given user i who rated movies j, sum up and average (r_{i,j} - r_j_avg)
Split data into multiples files for speed.
- one file for each movie? one file for each user?

Name		Name	Last commit message	Last commit date
Latest commit History 20 Commits
.gitignore		.gitignore
README.md		README.md
calculate_feature_errors.cpp		calculate_feature_errors.cpp
calculate_ratings.py		calculate_ratings.py
database_insert.py		database_insert.py
linear_regression.py		linear_regression.py
main.py		main.py
movie_features.dta.bak		movie_features.dta.bak
submission_svd.dta.bak		submission_svd.dta.bak
svd.cpp		svd.cpp
svd.py		svd.py
svd2.cpp		svd2.cpp
svd2.py		svd2.py
svd_paper.py		svd_paper.py
svd_test.py		svd_test.py
svm.py		svm.py
user_features.dta.bak		user_features.dta.bak

Provide feedback