The goal of this project (conducted in collaboration with the NUKKAI company) is to investigate how to build interpretable explanations for predictions of a credit scoring model learned from a real-world dataset provided by a microfinance institution.