近年来我国老龄化问题日益严峻，养老供给服务紧缺，养老社区发展空间大。本文以保险养老社区为例，阐述保险机构参与养老社区建设，不仅整合了健养服务产业链资源，建立覆盖全生命周期的保险保障与健养产业间增效提质机制，提升保单销售额，增加保费收入来源，优化资产负债久期匹配，提升投资收益率，同时养老社区获得保险机构引流提升入住率，实现双向赋能。对保险系养老社区入住率的关键影响因素进行实证研究，探索提升入住率的有效手段，有利于我国成功建设养老社区、提升保险养老社区的经济效益和社会效益，这对于整个社会、保险行业、养老产业具有重要理论和实践意义。 当前关于养老社区入住率影响因素的文献，主要为案例研究和定性研究，鲜少有针对保险系养老社区的相关定量研究。本文首先基于文献研究，选取引用率最高的影响因子作为定量模型参数，但这些参数不够全面且并不包括保险系养老社区的特征因子，因此本研究结合专家问询新增适用于我国保险养老社区实际情况的模型参数，并设计李克特五级量表，根据专家打分结果进行信度和效度检验后，科学筛选量化模型影响因子参数。然后基于调研获得的45个养老社区630条实际数据，训练随机森林模型，得出重要性权重始终较大的影响因子为：保险养老社区所提供的服务种类、开业年限、周边5公里内医疗配套数。再次，通过回归模型验证随机森林模型结论的有效性，对影响因子逐个进行分析，并输出入住率预测公式。最后，通过实验项目验证入住率预测结果与实际情况相符，进一步证明参数选择和模型结论的科学性和有效性，并根据模型结论针对性提出提升入住率的建议：一是加强医疗保障，做好医养结合；二是提供更多定制化服务，提高社区管理水平；三是加强品牌管理，维护养老社区长期稳定运营。 本文对标现有文献，依托入住率不高的实际问题，补充了保险系养老社区的入住率影响因素，并创新了入住率影响因素的定量研究方法，验证了模型与问题的契合度，以及模型结论的有效性，为提升保险养老社区入住率和竞争优势提供了可靠的理论依据和实证参考。关键词：养老社区；入住率；随机森林模型；回归模型

In recent years, China has been facing an increasingly severe aging problem, characterized by a shortage of elderly care services and significant potential for development in the elderly community sector. The participation of insurance institutions in the construction of elderly communities not only integrates resources within the healthcare and wellness service industry but also establishes mechanisms to enhance efficiency and quality between insurance protection and the healthcare industry, covering the entire lifecycle. This involvement aims to increase policy sales, diversify income sources, optimize asset-liability duration matching, and improve investment returns. Simultaneously, insurance institutions contribute to boosting the occupancy rate of elderly communities, thereby achieving mutual empowerment. This research empirically studies the key factors influencing the occupancy rate of insurance-based elderly communities, exploring effective strategies to enhance the occupancy rate. The findings will be beneficial for the successful development of elderly communities and the improvement of economic and social benefits in insurance-based elderly care. This research holds significant theoretical and practical implications for the whole society, the insurance industry, and the elderly care sector. The existing literature on factors influencing the occupancy rate of elderly communities primarily consists of case studies and qualitative research, with limited quantitative studies specifically focusing on insurance-based elderly communities. This study begins by reviewing the literature and selecting the most frequently cited factors as quantitative model parameters. However, these parameters are not comprehensive enough and do not encompass the unique characteristics of insurance-based elderly communities. Therefore, this research combines expert consultations to identify model parameters that are applicable to the actual conditions of insurance-based elderly communities in China. A Likert five-point scale is designed, and after conducting reliability and validity tests based on expert ratings, the model‘s quantitative influence factors are scientifically selected. Subsequently, using 630 actual data points collected from 45 surveyed elderly communities, a random forest model is trained to determine the importance weights of the influencing factors. The key factors consistently identified as having significant importance weights in the random forest model are the types of services provided by insurance-based elderly communities, the years of operation, and the number of medical facilities within a 5-kilometer radius. Furthermore, the regression model is employed to validate the effectiveness of the random forest model, analyze each influencing factor individually, and derive a formula for predicting the occupancy rate. Finally, the predicted occupancy rate is compared with actual data from experimental projects to further validate the scientific and effective selection of parameters and model conclusions. Based on the model findings, targeted recommendations are proposed to enhance the occupancy rate: strengthening healthcare support and integrating medical and elderly care services, providing more customized services to improve community management, and enhancing brand management to ensure long-term stable operation of elderly communities. This study builds upon existing literature and addresses the issue of low occupancy rates in insurance-based elderly communities. It supplements the current knowledge by identifying additional factors that influence the occupancy rate and introduces an innovative quantitative research method to analyze these factors. The study validates the model‘s alignment with the problem at hand and demonstrates the effectiveness of its conclusions. It provides a reliable theoretical basis and empirical reference for improving the occupancy rate and enhancing the competitive advantage of elderly communities.Keywords: retirement community; occupancy rate; random forest model; regression model