EquiPy:

In certain regions of Canada and the United States, discrimination based on multiple sensitive attributes, such as race and gender, is prohibited for insurance pricing. While approaches have been proposed to address these biases in predictive systems, they often struggle to offer a clear and accurate pathway to fairness, especially when multiple sensitive variables are involved. In this context, I will introduce EquiPy, a new open-source Python package implementing sequential fairness across multiple sensitive attributes via Optimal Transport. EquiPy will be demonstrated through an illustrative example using insurance data.