Risk Classification by Fuzzy Inference

Traditionally, policyholders in life insurance are classified in simple mortality tables, most often according to only a few risk characteristics. Instead of a risk classification according to the numerical rating system, this article describes how to classify by using a fuzzy inference methodology. By defining risk factors as fuzzy sets, it is shown that an insurer can utilize multiple prognostic factors that are imprecise and vague. The presented fuzzy risk classification provides a more realistic way of modeling mortality risks since it allows for compensations and interactions between multiple risk factors.