Abstract: The trajectory of car-haling is a type of spatiotemporal big data that can capture the characteristics of citizens’ mobility behavior. This paper constructs a travel complex network for car-hailing service from the perspective of network analysis and explores the spatiotemporal characteristics of mobility behavior, which is important for urban planning, route recommended, traffic optimization and so on. Based on the historical order data of DiDi service in Chengdu in November 2016, the travel complex network is constructed by dividing the main urban area into 400 urban grids and developing an OD matrix of car-hailing. This paper provides an in-depth analysis of the complex travel network structure, including centrality measures and the importance of mobility relationship, and develops multiple measurements to capture the spatiotemporal characteristics of mobility. The results show that the travel network for car-haling in Chengdu conforms to small-world network. Typical spatial differentiation characteristics, the distribution of node centrality, and the importance of mobility relationship follow truncated power-law functions. Besides, the imbalance of mobility behavior, the uncertainty of travel location distribution, and the imbalance of flow direction in the travel network have clear spatiotemporal evolution patterns. Moreover, for the current travel network, the density of travel volume which represents the travel scale will have a positive effect on these mobility characteristics.

Key words: car-hailing, mobility behavior, complex network, spatiotemporal big data, DiDi

摘要： 网约车轨迹是一类能够较好地反映市民流动行为的时空大数据。构建面向市民使用网约车的出行复杂网络，从网络的视角进行分析，进而挖掘网约车流动行为的时空特征，对于城市建设、路线推荐、道路优化等均有重要意义。基于滴滴出行的2016年11月份成都市内的历史订单数据，将主城区划分为400个城市网格，建立反映网约车流动的OD矩阵，构建出行复杂网络。对网络的复杂结构、各种中心度以及流动关系的重要性进行了深入分析，同时构建多个测度指标，度量了出行网络中网约车流动行为的时空特征。分析结果表明：成都市网约车的出行网络符合小世界理论，具有典型的空间分异特征，节点中心性和流动关系重要性的分布呈现显著的幂律形式；出行网络中流动行为的不平衡性、出行与到达位置分布的不确定性、流动方向的不平衡性均有较为明显的时空演变规律，且对于当前的出行网络，代表出行规模的出行量密度会正向影响这些流动特征。

关键词: 网约车, 流动行为, 复杂网络, 时空大数据, 滴滴