#Movie recommendation

• IO.py
• MovieClass.py
• RatingClass.py
• UserClass.py

• analyzeMovDB.py
  • analyses based on existing movies/rating/users
  • extract some interesting information such as
    • users' gender and corresponding rating
    • users' age
    • etc.
  • get highest score movies (with more than 100 ratings)
• rateMovies.py
  • for a new user, ask to rate some movies (highest score)
  • store rating & user information to new files
• recommend.py
  • use a specific user's rating and global database
  • choose the best model possible
  • recommend suitable movies for the user. Two modes possible:
    • "OFFLINE" mode: recommend for all users
    • "ONLINE" mode: recommend for user with Id 6041 (newly added)

Using (Weighted) Alternating Least Squares

Data: Movielens http://grouplens.org/datasets/movielens/ 1 million ratings, 6000 users on 4000 movies. Released 2/2003.

suggested by:
  http://www.analyticsvidhya.com/blog/2014/11/data-science-projects-learn/

Example of analyses inspired from http://bugra.github.io/work/notes/2014-04-19/alternating-least-squares-method-for-collaborative-filtering/ /!\ Data formats are different.