Agent-based simulation (ABS) models are potent tools for analyzing complex systems. However, understanding and validating ABS models can be a significant challenge. To address this challenge, cutting-edge data-driven techniques offer sophisticated capabilities for analyzing the outcomes of ABS models. One such technique is process mining, which encompasses a range of methods for discovering, monitoring, and enhancing processes by extracting knowledge from event logs. However, applying process mining to event logs derived from ABSs is not trivial, and deriving meaningful insights from the resulting process models adds an additional layer of complexity. Although process mining is invaluable in extracting insights from ABS models, there is a lack of comprehensive methodological guidance for its application in ABS evaluation in the research landscape. In this paper, we propose a methodology, based on the CRoss-Industry Standard Process for Data Mining (CRISP-DM) methodology, to assess ABS models using process mining techniques. We incorporate process mining techniques into the stages of the CRISP-DM methodology, facilitating the analysis of ABS model behaviors and their underlying processes. We demonstrate our methodology using an established agent-based model, Schelling model of segregation. Our results show that our proposed methodology can effectively assess ABS models through produced event logs, potentially paving the way for enhanced agent-based model validity and more insightful decision-making.

中文翻译：

---

基于 CRISP-DM 的方法，用于使用过程挖掘评估基于代理的仿真模型

基于代理的仿真 (ABS) 模型是分析复杂系统的有力工具。然而，理解和验证 ABS 模型可能是一项重大挑战。为了应对这一挑战，尖端的数据驱动技术提供了分析 ABS 模型结果的复杂功能。其中一种技术是流程挖掘，它包含一系列通过从事件日志中提取知识来发现、监控和增强流程的方法。然而，将流程挖掘应用于源自 ABS 的事件日志并非易事，并且从生成的流程模型中获取有意义的见解又增加了一层复杂性。尽管过程挖掘对于从 ABS 模型中提取见解非常有价值，但在研究领域中，其在 ABS 评估中的应用缺乏全面的方法指导。在本文中，我们提出了一种基于 CRoss 行业数据挖掘标准流程 (CRISP-DM) 的方法，使用流程挖掘技术评估 ABS 模型。我们将流程挖掘技术纳入 CRISP-DM 方法的各个阶段，促进 ABS 模型行为及其底层流程的分析。我们使用已建立的基于代理的模型（谢林隔离模型）来演示我们的方法。我们的结果表明，我们提出的方法可以通过生成的事件日志有效评估 ABS 模型，这可能为增强基于代理的模型有效性和更富有洞察力的决策铺平道路。