
科学技术与工程

中国技术经济学会
基于机器学习的社区建成环境对老年人步行时间的影响
朱震军 , 何展鹏 , 俞佳雯 , 韩吉 , 李晴
科学技术与工程 ›› 2025, Vol. 25 ›› Issue (18) :7832-7842
基于机器学习的社区建成环境对老年人步行时间的影响
The Impact of Machine Learning-based Community Built Environment on Walking Time of the Elderly
为探究社区建成环境对老年人步行时间的影响,考虑到老年群体的性别差异,构建CatBoost模型并结合SHAP解释方法(Shapley additive explanations, SHAP),对比分析社区建成环境与不同性别老年人步行时间的特征相对重要度及非线性关系。研究结果表明:相较于个人社会经济属性变量,社区建成环境变量对老年人步行时间影响更显著,但对不同性别老年人的影响效应有所差异。其中,相较于老年男性,老年女性更关注与社交需求密切的建成环境变量,如归一化植被指数(normalized difference vegetation index, NDVI)、人口密度等。而老年男性的步行时间与个人社会经济属性联系更加紧密,步行时间多与交通便捷性、出行效率相关。
To explore the impact of the community built environment on the walking time of elderly people, and considering gender differences among the elderly group, a CatBoost model was constructed and the SHAP (Shapley additive explanations) explanation method was integrated. The relative importance and nonlinear relationships of the community built environment features with the walking time of elderly people of different genders were comparatively analyzed. The study findings indicate that the community built environment variables have a more significant influence on the walking time of the elderly compared to personal socioeconomic attributes. However, the impact varies between genders. Compared to elderly males, elderly females pay more attention to built environment variables closely related to social needs, such as NDVI and population density. In contrast, the walking time of elderly males is more closely associated with personal socioeconomic attributes, often linked to transportation convenience and travel efficiency.
社区建成环境 / 老年人 / 步行时间 / CatBoost模型 / SHAP解释方法
community built environment / elderly people / walking time / CatBoost model / SHAP explanation method
/
| 〈 |
|
〉 |