基于XGBoost-SHAP框架的我国农村居民生活质量测度与影响因素研究

Research on Measuring and Influencing Factors of Rural Residents’ Quality of Life in China Based on XGBoost SHAP Framework

摘要: 本文基于全国性微观调查数据CFPS (2020)，在多维生活质量框架下构建涵盖经济能力、教育水平、健康状况、社会支持、公共服务、数字参与与主观满意度等多个领域的农村居民生活质量指标体系。通过熵权法与因子分析分别构建客观与主观生活质量指数，并采用XGBoost模型对多维指标进行非线性融合，形成综合生活质量指数。同时，引入SHAP方法对模型进行解释，识别不同指标对生活质量的边际贡献及其影响机制。研究结果表明：第一，农村居民整体生活质量水平偏低，呈现显著右偏分布，中低生活质量群体占比大，整体表现出“低位集中、尾部有限”的结构特征。第二，从区域视角看，东、中、西部的生活质量水平差异并不显著，三大区域分布高度重叠，区域经济发展的差距尚未充分传导至农村家庭生活质量层面。第三，农村生活质量呈现“公共服务质量–社会支持网络–经济能力”三元结构，其中教育公共服务和社会互助网络贡献度最高，收入与教育能力发挥重要基础作用，而健康状况、医疗体验和数字参与的边际贡献相对较弱。研究揭示了农村生活质量的结构性特征与关键驱动机制，为精准改善农村居民福祉和优化公共服务供给提供了重要政策参考。

Abstract: Drawing on microdata from the China Family Panel Studies (CFPS 2020), this study measures the quality of life of rural residents in China using a multidimensional framework that incorporates economic capacity, education, health status, social support, public services, digital participation, and subjective life satisfaction. To capture both objective conditions and subjective perceptions, the entropy weight method and factor analysis are first used to construct objective and subjective quality-of-life indices. These indicators are subsequently integrated through the XGBoost algorithm to generate a comprehensive index, and the SHAP method is employed to interpret the model and quantify the marginal contribution of each factor. The results show that the overall quality of life of rural residents remains relatively low and displays a pronounced right-skewed distribution, indicating that a large share of the population is concentrated at lower levels of well-being, with a relatively limited upper tail. Regional comparisons suggest that differences among eastern, central, and western China are not statistically significant, and the distributions across the three regions largely overlap. This implies that macro-level regional economic disparities have not yet been fully translated into differences in rural household living conditions. Further analysis indicates that rural quality of life is primarily shaped by a structural combination of public service quality, social support networks, and economic capacity. Among these factors, educational public services and social support show the strongest contributions, while income and educational attainment provide an important foundational basis for improvement. In contrast, health status, healthcare experience, and digital participation exhibit relatively weaker marginal contributions. These findings provide important policy implications for improving rural well-being and optimizing the provision of public services.