A novel predict-then-optimize method for sustainable bike-sharing management: a data-driven study in China

Area of Science:

• Operations Management
• Urban Transportation Systems
• Sustainable Logistics

Background:

• Post-pandemic recovery sees increased demand for low-carbon transportation like bike-sharing.
• Significant challenges exist in optimizing the reallocation of shared bikes across urban networks.
• Inefficient reallocation leads to transportation mismatches and resource waste.

Purpose of the Study:

• To develop and evaluate a novel predict-then-optimize method for efficient bike-sharing operations.
• To minimize operational costs and enhance transportation efficiency in urban bike-sharing systems.
• To facilitate the sustainable development of urban transportation networks.

Main Methods:

• A data-driven robust optimization model to predict demand surplus at each location.
• A branch-and-price algorithm to determine optimal bike reallocation routes based on predictions.
• Deployment and testing of the integrated predict-then-optimize method on a real-world Chinese bike-sharing network.

Main Results:

• The method successfully predicted demand surplus using historical data.
• Optimal reallocation schedules were derived, minimizing operational costs.
• The branch-and-price algorithm identified efficient routes for bike assignment.
• Significant cost savings and reduced resource waste were demonstrated.

Conclusions:

• The novel predict-then-optimize method effectively addresses bike-sharing reallocation challenges.
• This approach offers a sustainable solution for urban transportation systems.
• The method has high potential for widespread adoption in bike-sharing networks globally.