Streamsight is an offline Reccomender Systems (RecSys) evaluation toolkit that respects a global timeline. The aim is to partition the data into different windows where data is incrementally released for the programmer to fit, train and submit predictions. This aims to provide a close simulation of an online setting when evaluating RecSys algorithms.
Clone the repository
git clone https://github.com/hiiamtzekean/streamsight
cd streamsightDependencies can be installed with uv for ease of management.
uv syncAlternatively, you may install dependencies locally with pip and venv
python3 -m venv venv
source venv/bin/activate
pip install -e .The dependencies are listed in pyproject.toml.
- We welcome all contributors, be it reporting an issue, or raising a pull request to fix an issue.
- When you make changes, rerun
pip install .to test your changes.
The documentation can be found here and repository on Github.
If you use this library in any part of your work, please cite the following papers:
Ng, T. K. (2024). Streamsight: a toolkit for offline evaluation of recommender systems. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/181114