Evaluating public transport performance to guide public transport planning: A unified prior method

Abstract

Public transport performance (PTP) depicts the efficiency of a city’s public transport. Evaluating PTP is the foundation to build a sustainable public transport system. However, extant methods are not suitable for evaluating PTP at the planning stage. This paper initiates a unified prior evaluation method that guides the planning of public transport without observational data. We represent the multimodal public transport system as a multigraph, where node function and connectivities are embedded as node property, planning specification and graph metrics are incorporated as the weight of the edge. By node convolution and node clustering, eight typical subgraphs are enumerated. Graph classification evaluates the PTP by identifying combinations of subgraphs within a region and motif analysis reveals how local transfer patterns contribute to PTP. The experiment results within Wuhan’s Third Ring Road show that our evaluation accuracy is up to 82%. We found that areas with high accessibility and equality of public transport facilities also have high PTP. From small scales (stations) to large scales (regions), PTP evolves from spatial homogeneity to heterogeneity. Using these features, urban planners can design an efficient public transport system with a point-to-surface effect and integrate ground and underground transport modes.