Mapping Borrowers’ Data: A Topological Approach to Credit Scoring

This study introduces a topological approach to credit scoring, leveraging the insights and tools of topological data analysis (TDA). It presents a conceptually straightforward yet theoretically robust topological scoring model, designed to complement established methodologies and enhance credit assessment through superior explainability and visualization. This topological approach elucidates the high-dimensional feature space of borrowers, visually representing credit quality across varied feature neighborhoods and exploring the connectedness between these areas. This not only fosters a deeper understanding of creditworthiness determinants but also provides a basis for targeted guidance to borrowers, helping them improve their positioning in credit neighborhoods. Furthermore, the authors validate the predictive performance of their model across diverse datasets, demonstrating its robustness and competitive standing against traditional and modern credit scoring methods.