Skip to content

Latest commit

 

History

History
14 lines (10 loc) · 598 Bytes

README.md

File metadata and controls

14 lines (10 loc) · 598 Bytes

Bayesian-Personalized-Ranking

One-class recommendation algorithm from implicit feedback

Description

For implicit feedback recommendation problem, One-class recommendation (BPR) often performs well for either classification (task 1) or regression (task 2). This project participated in Kaggle competition https://www.kaggle.com/c/cse158-258-fa17-visit-prediction, and won the 1st place.

Environment

Python 3.7

Data

Training data should be downloaded by yourself, from URL implied in description txt.

Metric

Measurement we use for task 1 is prediction accuracy, and RMSE for task 2