Exploring spatial association between residential and commercial urban spaces: A machine learning approach using taxi trajectory data

Abstract

Human mobility datasets, such as traffic flow data, reveal the connections between urban spaces. A novel framework is proposed to explore the spatial association between urban commercial and residential spaces via consumption travel flows in Shanghai. A social network analysis and a community detection method are employed using taxi trajectory data during the daytime to validate the framework. The machine learning-based approach, such as the community detection method, can overcome the limitation regarding spatial uncertainty and spatial effects. The empirical findings suggest that people's commercial activities are sensitive to the power of accessible commercial centers and travel distances. The high-level commercial centers would contribute to the monocentric structure in the outer urban region based on consumption flows. In the central urban region, increasing the number of high-level commercial centers and making the powers of commercial centers hierarchical can contribute to a polycentric mobility pattern of people's consumption. This research contributes to the literature by providing a novel framework to model, analyze and visualize people's mobility based on the trajectory big data, which is promising in future urban research.