北京大学学报自然科学版 ›› 2020, Vol. 56 ›› Issue (3): 518-530.DOI: 10.13209/j.0479-8023.2020.035

上一篇    下一篇

基于手机信令数据的特大城市人口时空分布及其社会经济属性估测——以北京市为例

海晓东1, 刘云舒2,3, 赵鹏军3,†, 张辉1   

  1. 1. 北京大学经济学院, 北京100871 2. 北京大学深圳研究生院, 深圳 518055 3. 北京大学城市与环境学院, 北京 100871
  • 收稿日期:2019-05-10 修回日期:2020-01-16 出版日期:2020-05-20 发布日期:2020-05-20
  • 通讯作者: 赵鹏军, E-mail: pengjun.zhao(at)pku.edu.cn
  • 基金资助:
    国家自然科学基金(41925003)和北京建筑大学未来城市设计高精尖创新中心项目(udc2018010921)资助

Using Mobile Phone Data to Estimate the Temporal-Spatial Distribution and Socioeconomic Attributes of Population in Megacities: A Case Study of Beijing

HAI Xiaodong1, LIU Yunshu2,3, ZHAO Pengjun3,†, ZHANG Hui1   

  1. 1. School of Economics, Peking University, Beijing 100871 2. Shenzhen Graduate School, Peking University, Shenzhen 518055 3. College of Urban and Environmental Sciences, Peking University, Beijing 100871
  • Received:2019-05-10 Revised:2020-01-16 Online:2020-05-20 Published:2020-05-20
  • Contact: ZHAO Pengjun, E-mail: pengjun.zhao(at)pku.edu.cn

PDF

摘要：

提出应用手机信令数据, 基于空间模式单元(Spatial Pattern Unit)进行人口动态分布估测和人口属性识别的方法, 并以北京为例开展实例研究。以手机信令数据为主, 结合大样本问卷调查数据和腾讯热力图数据, 对人口布局进行分时段估测, 分析人口分布的时空间动态特征; 采用大样本问卷调查数据, 以人口社会经济属性和通勤出行特征等关键指标, 对调查的种子空间单元进行模式分类和识别, 运用机器学习的方法进行全域地域空间的人口属性估测识别, 最后对估测结果进行对比和验证。所提方法和研究结果可以为监测人口布局动态、针对人口属性布局商业服务和合理规划城市设施等提供决策支撑。

关键词: 人口时空分布, 人口时空分布, 人口属性估测, 人口属性估测, 动态监测, 动态监测, 机器学习, 机器学习, 手机信令数据, 手机信令数据

Abstract:

This study proposes a technique to identify the temporal-spatial distribution and socioeconomic attributes of population by using mobile phone data. This technique has a fine geographic scale, which is called as Spatial Pattern Unit. The study uses Beijing as a case and conducts an empirical application of the technique. Firstly, it investigates the temporal-spatial distribution of population in Beijing by using multiple data sources, including mobile phone data, travel survey data and heat map data. Secondly, it classifies the spatial pattern unit into different categories in terms of socioeconomic attributes of population and travel behavior features. Thirdly, it applies machine learning approach to estimate socioeconomic attributes of population for all spatial pattern units. Finally, it compares and verifies the results of analysis. The approaches and findings would be valuable to monitoring population distribution, locating business services and planning urban infrastructure.

Key words: temporal-spatial distribution of population, temporal-spatial distribution of population, estimation of socioeconomic attributes of population, estimation of socioeconomic attributes of population, dynamic monitoring, dynamic monitoring, machine learning, machine learning, mobile phone data, mobile phone data

版权所有 © 《北京大学学报（自然科学版）》编辑部
技术支持：北京玛格泰克科技发展有限公司