Agent-Based Model (ABM) validation considerations

Abstract

This paper describes the use of validation methods in model building. We address issues associated with the increasing complexity of models that is in part a response to the growing popularity of Agent-Based Models (ABM), commonly used to study cognitive, natural, and social phenomena. The first section of this manuscript discusses model categories and attributes. The second section discusses the stages of validating a simulation model: verification, validity, and sensitivity analysis. The third section presents specific validation approaches, with an emphasis on six specific tests that are described in detail. The final section summarizes the goals of model validation and modeling.