Demand Prediction and Influencing Factors Analysis of Shared Bicycles near Public Transportation Stations based on GCN

Abstract

Traffic prediction is a pivotal technology in the development of intelligent transportation systems. Real-time and accurate traffic prediction holds significance in route planning, resource allocation, and improving travel efficiency. To enhance the convenience of shared bicycle travel near public transportation hubs and optimize shared bicycle deployment strategies, this paper proposes the utilization of a graph convolutional neural network prediction model. Initially, shared bicycle travel data is scrutinized for spatiotemporal characteristics. Subsequently, an adjacency matrix is constructed based on the spatial attributes of the data to establish the graph convolutional network. Lastly, global spatial autocorrelation analysis and a coupled coordination model are integrated for validation and augmentation. This model is employed to predict shared bicycle travel patterns near public transportation hubs in the Xiamen Island area. Experimental results demonstrate a high level of precision, underscoring the model's effectiveness in providing valuable guidance for shared bicycle travel. The effectiveness of this method can be further verified in subsequent empirical studies.