An Application of a Fuzzy Multi-criteria Decision Making Process for Explainable Machine Learning in the Actuarial Context
摘要
There is an ever-increasing need for system interpretability in artificial intelligence and machine learning. Explanation, or system interpretability, has always been necessary in applications where critical decisions, for example, in actuarial context applications, need to be made. This work presents a case study of applying Fuzzy Multi-Criteria Decision-Making (MCDM) models for solving actuarial problems using machine learning methods. The used explainable framework includes a Fuzzy Inference System paired with a modified MCDM-based model to obtain a rank of relevant variables both in global and local decisions.