Exploring spatial heterogeneity in the impact of built environment on taxi ridership using multiscale geographically weighted regression

Pengyu Zhu,Jiarong Li,Kailai Wang,Jie Huang

DOI: https://doi.org/10.1007/s11116-023-10393-1

IF: 4.814

2023-05-17

Transportation

Abstract:Due to its flexibility and door-to-door service, taxis are an integral part of the urban transportation system. They have become an essential solution to the first/last mile problem. Even though much research has been conducted on the effects of built environment variables on taxi passengers' travel behaviors, few have accounted for the spatial heterogeneity embedded in multiscale spatial processes. This study applies multiscale geographically weighted regression (MGWR) to investigate the associations between taxi ridership and spatial contexts to address the gaps. The MGWR considerably improves modeling fit compared to the global OLS model by capturing the spatially varying processes at different scales. The results demonstrate the existence of strong spatial non-stationarity in the various built environment factors affecting the spatial distribution of taxi pick-ups and drop-offs. Specifically, increased residential density induces more taxi demand in areas with less access to public transportation than their surrounding units. Increasing bus coverage where bus coverage is relatively low may attract more commuters to adopt taxi plus bus mode for commuting. Road network density has a more substantial effect on taxi ridership in the south end of the city than in the north. The former is characterized by lower road density. This study reveals the complex relationships between the built environment and the distribution of taxi ridership at different spatial scales and provides valuable insights for transport planning, taxi resource allocation and urban governance.

transportation,transportation science & technology,engineering, civil