yyf1997keykeyup/recommend_system_for_movielens

This project is created in cooperation by Chen Qian, Yufeng Yan, Yinuo Peng, Yu Xie for UVa CS 6316 Machine Learning course.

Recommendation System for MovieLens Dataset

Authors

This project is created in cooperation by Chen Qian, Yufeng Yan, Yinuo Peng, Yu Xie for UVa CS 6316 Machine Learning course.

Python Environment

• python: 3.0 - 3.8

How to install and run

# install PyYAML lib
pip install -r requirements.txt

# (optional) modify config.yml file as you want
vim config.yml

# run the main.py
python main.py

console logging output example:

[INFO] model_name: UserCF-IIF, dataset: ml-100k, test_ratio: 0.100000
[SUCCESS] dataset has been loaded
[SUCCESS] split data has been split to a test set and a train set
[INFO] the ratio of train set = 89961
[INFO] the ratio of test set = 10039
[INFO] start UserBasedCF
[INFO] No model saved in model dir
[INFO] Training a new model...
[INFO] building movie-users inverse table
[SUCCESS] building movie-users inverse table success.
[INFO] the number of all movies = 1671
[INFO] calculating users co-rated movies similarity
[INFO] 0 steps, 0.01 seconds have spent
[INFO] 500 steps, 7.86 seconds have spent
[INFO] 1000 steps, 9.55 seconds have spent
[INFO] 1500 steps, 9.74 seconds have spent
[SUCCESS] Got user co-rated movies similarity matrix success, spent 9.739250 seconds
[INFO] calculating user-user similarity matrix
[INFO] 0 steps, 0.00 seconds have spent
[INFO] 500 steps, 0.49 seconds have spent
[SUCCESS] Got user-user similarity matrix success, spent 0.922652 seconds
[Success] A new model has been trained
[SUCCESS] The new model has saved
[INFO] start to recommend based on the recommend_user_list
[SUCCESS] For user whose id = 1, The id of the recommended movies:
['168', '423', '433', '4', '603', '318', '313', '474', '568', '288']
[SUCCESS] For user whose id = 2, The id of the recommended movies:
['124', '303', '750', '515', '248', '9', '237', '181', '744', '137']
[SUCCESS] For user whose id = 3, The id of the recommended movies:
['313', '286', '269', '751', '315', '301', '748', '750', '895', '310']
[SUCCESS] For user whose id = 100, The id of the recommended movies:
['748', '332', '682', '307', '339', '327', '343', '245', '304', '678']
[SUCCESS] For user whose id = 500, The id of the recommended movies:
['64', '132', '318', '427', '12', '732', '7', '14', '173', '22']
[SUCCESS] For user whose id = 900, The id of the recommended movies:
['127', '50', '479', '98', '286', '276', '179', '174', '515', '496']
[INFO] start testing
[INFO] 0 steps, 0.01 seconds have spent..
[INFO] 500 steps, 1.86 seconds have spent..
[SUCCESS] Test recommendation system success, spent 3.154643 seconds
[SUCCESS] precision=0.18876	 recall=0.17731	 coverage=0.24955	 popularity=5.42733	
[INFO] total 14.67 seconds have spent

What's More

Some parts of the code are based algorithms on the book -- 'Recommended system practice' written by Liang Xiang.