A Locational Demand Model for Bike-Sharing

• 成果类型: Article
• 署名作者: Xu, Ang; Yan, Chiwei; Goh, Chong Yang; Jaillet, Patrick
• 署名单位: University of California System; University of California Berkeley; Uber Technologies, Inc.; Massachusetts Institute of Technology (MIT)
• 刊物名称: M&SOM-MANUFACTURING & SERVICE OPERATIONS MANAGEMENT
• ISSN/ISSBN: 1523-4614
• DOI: 10.1287/msom.2023.0306
• 发表日期: 2025
• 关键词: locational demand model; bike-sharing; expectation-maximization; location discovery

摘要

Problem definition: Micromobility systems (bike-sharing or scooter-sharing) have been widely adopted across the globe as a sustainable mode of urban transportation. To efficiently plan, operate, and monitor such systems, it is crucial to understand the underlying rider demand-where riders come from and the rates of arrivals into the service area. They serve as key inputs for downstream decisions, including capacity planning, location optimization, and rebalancing. Estimating rider demand is nontrivial as most systems only keep track of trip data, which are a biased representation of the underlying demand. Methodology/results: We develop a locational demand model to estimate rider demand only using trip and vehicle status data. We establish conditions under which our model is identifiable. In addition, we devise an expectation-maximization (EM) algorithm for efficient estimation with closed-form updates on location weights. To scale the estimation procedures, this EM algorithm is complemented with a location-discovery procedure that gradually adds new locations in the service region with large improvements to the likelihood. Experiments using both synthetic data and real data from a dockless bikesharing system in the Seattle area demonstrate the accuracy and scalability of the model and its estimation algorithm. Managerial implications: Our theoretical results shed light on the quality of the estimates and guide the practical usage of this locational demand model. The model and its estimation algorithm equip municipal agencies and fleet operators with tools to effectively monitor service levels using daily operational data and assess demand shifts because of capacity changes at specific locations.