Personal Credit Risk Measurement: Bilateral Antibody Artificial Immune Probability Model

Abstract

This article presents a credit risk model for measuring personal default probability by introducing the immunity algorithm. Compared with logistic regression model with Receiver Operator Curve (ROC) test, the model which under the theoretic framework of bilateral antibody artificial immunity appears to be more sensitive to sample data and competent in prediction. The most distinguishing feature owned by this model is it could evolve if trained, and this makes it intelligent and dynamic. Furthermore, it could be implemented not only in predicting individual default probability of commercial banks' clients, but also in measuring personal credit character for other public services.