Dynamic Scheduling Strategy Based on Demand Prediction of Shared Bike
摘要
This paper aims to address the issue of imbalanced supply and demand in shared bikes by proposing the concept of ‘red packet bike’, utilizing the open-source Beijing Shared Bike Dataset. The Temporal Graph Convolutional Network (T-GCN) is selected to predict the demand for shared bikes based on the analysis of spatio-temporal correlations in the order data. Additionally, a user incentive scheduling algorithm is designed using the breadth-first algorithm (BFS) and presented in the form of distributing the red packet bikes, thereby delegating the scheduling problem to the users to solve.