本文针对传统的三支决策方法面临的三个关键挑战进行了深入研究。首先，传统方法需要明确定义主观阈值，这限制了其在不同场景下的适用性。其次，传统方法在信息融合方面缺乏强解释性，难以提供清晰的决策依据。因此，本文利用复模糊集构建可调多粒度复模糊信息系统，并结合基于决策的三支决策方法客观获取分类阈值，构造了可调多粒度复模糊概率粗糙集。最后，决策者常常受制于有限理性，这在实际决策中不可避免。基于此，本文在行为决策理论下，将决策者在决策过程中心理因素对结果的影响展现出来。综上所述，本文提出了一种基于复模糊集和后悔理论的多粒度三支决策模型，旨在解决复杂多属性群决策问题。进一步地，本文利用共享单车租赁数据集进行了详细的实验分析，验证了该模型在求解复模糊环境下蕴含决策风险和有限理性的决策问题时的可行性和有效性。同时，在容错能力和专家经验所发挥的效能上，都获得了较大提升。

This paper embarks on an extensive exploration of the challenges inherent in traditional three-way decision-making methodologies. First, these conventional approaches necessitate precise subjective threshold definitions, rendering them less adaptable across diverse scenarios. Second, their interpretability in information fusion is often lacking, resulting in ambiguous decision criteria. To address these issues, this paper introduces a novel approach that leverages complex fuzzy sets to establish an adjustable multi-granularity complex fuzzy information system. It integrates decision-based three-way decision techniques to derive classification thresholds objectively, yielding an adjustable multi-granularity complex fuzzy probabilistic rough set. Furthermore, the paper acknowledges the constraints of bounded rationality that decision-makers encounter in real-world scenarios. Within the framework of behavioral decision theory, this work provides a more comprehensive depiction of how psychological factors influence decision outcomes. In essence, this paper proposes a multi-granularity three-way decision model grounded in complex fuzzy sets and regret theory, designed to tackle complex multi-attribute group decision-making problems. To validate the model's effectiveness in addressing decision challenges related to decision risk and bounded rationality within a complex fuzzy environment, the paper conducts an in-depth experimental analysis using a shared bicycle rental dataset. The results underscore the model's feasibility and efficacy, particularly in enhancing fault tolerance and leveraging expert experience.