A dual-metric public bike system performance assessment: the case of Taichung YouBike

Research into public bike systems (PBSs) has focused mainly on either user or operator perspectives. This study bridges the gap by introducing an integrated dual-metric model combining level of service (LoS) and level of utilisation (LoU). Using Taichung’s YouBike as a case study, data from 911 stations over 40 days were analysed, including 10.5 million bike and dock availability records for LoS and 1.4 million rental transaction records for LoU. LoS was measured by station reliability and LoU by station turnover rates. Stations were classified into G1 (high LoS and LoU), G2 (medium-low LoS, high LoU), G3 (high-medium LoS, low LoU), and G4 (low LoS and LoU). Geographic, socioeconomic, built environment and land-use zoning data around each station were collected from various open sources. The random forest technique was used to identify attributes influencing the group types. The findings revealed PBS trip characteristics, improvement strategies for G4 stations and optimal station density and size. This study offers a practical and adaptable framework for assessing PBS performance.