pytest-cases: a Python package for reproducible AI results (among others)

`pytest-cases` is a Python package leveraging the widely popular `pytest` library (Krekel et al., 2004) to design tests and benchmarks where test data, tested code, and evaluation protocols are strictly separated. With `pytest-cases` it has never been so easy for researchers to create reproducible results tables !

The API for `pytest-cases` is as minimal as possible, so that each concept (test data, algorithms candidates, evaluation protocols) is a readable python function ; this is in line with the design philosophy of `pytest` itself.