Multi-armed single/multi-agent bandits Python package.
Bandpy aims to provide all classical agents and controllers on a various synthetic and real data environnments to ease benchmark for research and R&D purposes.
- Official source code repo: https://github.com/hcherkaoui/bandpy
In order install the package, run::
pip install -r requirements.txt
pip install -e .
To validate the installation, run::
cd examples/1_illustrations
python 0_demo_lin_ucb.py # to play around examples
The required dependencies to use the software are:
- Matplotlib (>=3.0.0)
- Numpy (>=1.10.0)
- Pandas (>=1.4.1)
- Scipy (>=1.8.0)
- Joblib (>=0.16.0)
- Scikit-Learn (>=1.0.2)
- Networkx (>=2.8.6)
- Gym (>=0.23.1)
- Wget (>=3.0)
- matrix_factorization
In order to launch the unit-tests, run the command::
pytest # run the unit-tests
In order to check the PEP 8 compliance level of the package, run the command::
flake8 --ignore=E501 --count bandpy