Abstract

Agent-based models can simulate interactions of complex systems that lead to emergent events. This capability enables the exploration of potential outcomes of different assumptions and scenarios, which can support the decision-making process for complex systems. However, identifying the optimal level of detail and granularity of agent-based models is challenging and highly related to the decisions they are intended to support. Detailed and granular models can incorporate more information and potentially provide a more realistic representation of an actual system. However, the more complex models require more time and resources to run and analyze, and their complexity can make the interpretation of simulation challenging. Conversely, simpler and more aggregated models are often easier to interpret and more efficient to run, though they may offer a less accurate representation of the original system. In this paper, we discuss the trade-offs between detailed and aggregated models and review the factors that influence the optimal level of detail and granularity.