Neutrosophic Extension of the Hybrid Logistic Model in Credit Scoring

Abstract

This paper offers a methodological proposal to augment the Hybrid Linear Logistic Model (HLLM) [1]. for applied credit rating by introducing a neutrosophic logic structure. The original model blends linear and nonlinear elements via variable-specific basis function expansions and has demonstrated validity and interpretability based on requirements set forth by Basel II. However, it still assumes the classical form of uncertainty. To fill this void, we propose a neutrosophic reformulation that allows each explanatory variable to be specified not just by a point value, but as a triplet (T, I, F), which represent degree of truth (T), indeterminacy (I) and falsity (F) associated with the provided information. A Neutrosophic Probability of Default (PDⁿ) that is more aligned to the uncertainty and variability in a customer’s financial behaviour, emerges from our proposal. The new model retains the functional organization of the HLLM, while increasingly its robustness and interpretability by incorporating weighted neutrosophic basis functions. This enhances the interpretative capacity of the credit rating system under uncertain conditions or with incomplete qualitative or contradictory information.