Metro station risk classification based on smart card data: A case study in Beijing

Abstract

As the hub of urban railway transit, metro stations portray the skeleton structure of the public transit network. This study proposes a method of station classification from the dual perspectives of network structure and passenger flow. Each perspective considers the two aspects, one is the characteristics of the node itself, such as degree and the entrance and exit ridership; another considers the characteristics of the influence of other nodes, such as betweenness centrality and passing flow. Among them, the importance index of passing flow is calculated by the PageRank algorithm. According to these characteristics, metro stations are classified by k-means clustering algorithm after dimensionality reduction. The case study is conducted through nearly five million records from 278 stations in Beijing. From the classification results, stations are divided into six categories. Qualitative and quantitative regulations are proposed to reduce the risk of high ridership stations and improve the operation efficiency for few ridership stations.