Conditional Weight of Evidence for Gradient-Boosted Credit Models: Estimation, Additivity and Its Relation to Adverse Action Reasons
Resumo indisponível. Consulte o paper original.
PAPER / ARXIV:2610.07888
Jonas Brunholm, Bjarne Højgaard, Thomas D. Nielsen, Orimar Sauri
RESUMO
This paper proposes a framework for modeling probability of default via functional data analysis. By representing a series of credit variables as functions, we investigate whether intra-monthly information improves default predictability in linear models. We show that variables such as available funds, utilization rate, and overdraft can all be derived from three quantities observed over time: account type, balance, and credit limit. Retaining these quantities as continuous-time processes yields a continuous faithful representation of the borrower. We assess the framework using functional logistic regression, which attains predictive performance on par with XGBoost while consistently outperforming logistic regression, balancing predictive performance with interpretability under strict regulation.
NO MESMO MAPA
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.