Recommendation-System

Overview

CF.py：Item-based collaborative filtering algorithm.

ICCF.py：Introduce item clustering into collaborative filtering algorithm.

SVD.py：Singular Value Decomposition algorithm.

data_manager.py：Data process related class.

multiProcess.py: Concurrently calculate Pearson Correlation Coefficient by Ratings and Euclidean Distance by Tag Genomes.

Utilizing the Project

Previously run multiProcess.py to calculate correlation coefficients and save them locally. Then use CF.py, ICCF.py or SVD.py to train different recommendation models.

Benchmark

RMSE comparison of different algorithms

Method	RMSE
MovieAvg	1.1162
Item-based CF	0.9770
SVD	0.9256
ICCF	0.9113

Precision and Recall comparison between ICCF and SVD

F-score comparsion between ICCF and SVD