Discovering Station Patterns of Urban Transit Network with Multisource Data: Empirical Evidence in Jinan, China

The various performances of buses at stations bring lots of difficulties for operators to manage them to improve the service quality. This paper proposes a data-driven framework to analyze the patterns of stations with network structure data, points of interest (POI) data and vehicle global positioning system (GPS) trajectory data. First, we build six indicators based on these data to measure the performance from station perspective. The results show that the number of POI around stations within 1 kilometer follows an exponential distribution. Moreover, the average headway and headway deviation of stations follow lognormal distributions. Second, we use agglomerative hierarchical clustering method to divided bus stations into different groups. Results indicate that the bus stations of Jinan could be divided into four groups with obvious characteristics. The findings could help operators to make exclusive strategies to manage bus systems.