How to quantify the travel ratio of urban public transport at a high spatial resolution? A novel computational framework with geospatial big data

Improving the travel ratio of public transportation (PTR) is important for realizing low-carbon transportation and sustainable city development. However, limited by data resolution and model accuracy, existing research rarely involves the spatially refined calculation of PTR and the quantitative analysis of its influencing factors. In this study, based on multi-source geospatial big data, we propose a novel computational framework to solve the above problems. Specifically, we first design a linear programming-based three-step method, which realizes the calculation of PTR at 500-meter grid-pair scale for the first time; secondly, we develop a Beta-binomial model for regression analysis, which improves by more than 50% compared with traditional generalized linear models. The case of Wangjing area in Beijing shows that: the overall PTR in Wangjing is only 16%, which is much lower than the official expectation (45%), and less than 20% of origin–destination (OD) pairs meet the standard; among the influencing factors, the travel duration gap between public transportation and private cars, walking distance, number of transfers, and residential parking density have significant negative effects on PTR. Finally, this paper provides an implication of the proposed computational framework, i.e., the accurate detection of public transportation (PT) supply–demand imbalance areas, which proves its great potential in refined transportation optimization and sustainable urban planning.