Movie recommendation engine written in C++ using the Netflix matrix factorization technique
Matrix factorization techniques for recommender systems
A rating dataset from MovieLens containing 100004 records is imported in the project.
It is located in data/ratings.csv
mkdir build
cd build
cmake ..
make
./src/recommender csvfile k eta lambda [iteration number]
Run the recommender on the ratings dataset:
./src/recommender ../data/ratings.csv 2 0.001 0.01
For fast test use small.csv file:
./src/recommender ../data/small.csv 2 0.04 0.0001 100000
Results are written into the results.txt text file.