Stepping on earth: A roadmap for data-driven agent-based modelling

Abstract

In the last few years an increasing number of more complicated Agent-Based Models have been designed with the aim of approaching reality more closely, usually by introducing more and more empirical data. In this paper we review different approaches that could be adopted: KISS, KIDS and a new one that lies between both. We then propose a new logic of simulation driven by empirical data. Because the emphasis on collected data affects the model’s design and initialisation, as well as its validation, we provide guidelines for data injection, flow diagrams of well-defined stages, and suggest the application of some Artificial Intelligence technologies.