A CRISP-DM-based methodology for assessing agent-based simulation models using process mining

Agent-based simulation (ABS) models are powerful tools for analyzing complex systems. However, understanding and validating ABS models can be challenging. Data-driven techniques, such as process mining, offer promising capabilities for addressing these challenges. Process mining enables the discovery, monitoring, and enhancement of processes by extracting insights from event logs. However, applying process mining to ABS-generated logs and interpreting the results is not trivial. Despite its potential, limited methodological guidance exists for using process mining in ABS evaluation. This paper proposes a methodology, grounded in the CRoss-Industry Standard Process for Data Mining (CRISP-DM), to assess ABS models via process mining. By integrating process mining techniques into the phases of CRISP-DM, we support the analysis of ABS behaviors and their underlying processes. We demonstrate our methodology using Schelling’s segregation model. Our results indicate that our proposed methodology effectively evaluates ABS models using event logs, enhancing model validity and supporting more informed decision-making.