Revealing job accessibility supported by public transit services using a graph convolutional network

Abstract

Encouraging the use of public transit is widely recognized as a viable solution to address urban transportation issues such as increasing congestion and pollution. Public transit is currently one of the most popular modes of transportation for commuting in metropolitan areas. Although numerous studies have developed methods to address job accessibility issues, overcrowding and long travel time issues in public transit commuting remain unresolved. Currently, the role of public transit in supporting job accessibility based on actual commuting demands has not been fully explored. This research aims to uncover the contribution of public transit to job accessibility from a network modeling perspective. Job accessibility and public transit services are evaluated and synthesized into an origin-destination (OD) network derived from real commuting demand extracted from big data. A graph convolutional network clustering method is applied to reveal the network characteristics of the public transit supportability for job accessibility. Urban function features are further analyzed to identify the potential improvement for poor job accessibility places. The results obtained through a case study in Shenzhen identify the regions where conflicts exist between job accessibility and the public transit supply. The analysis reveals specific deficiencies in job accessibility, offering a deeper understanding of the root causes of commuting issues. Possible solutions for improving public transit and land use to enhance job accessibility are discussed based on the empirical findings. The insights gained from this research are valuable for urban planning and transportation improvements aimed at addressing urban commuting challenges.