Dynamic Spatial-Temporal Graph Attention Graph Convolutional Network for Short-Term Traffic Flow Forecasting

The application of graph convolutional network in short-term traffic flow forecasting of road network has effectively improved the prediction accuracy. The key point of this method is to construct the Laplacian matrix through extracting spatial features among nodes of the road network. However, most available methods mainly rely on the spatial distance among nodes to construct Laplacian matrix, then optimized the Laplacian matrix by other methods, which limits the wide application of the model. In this paper, we propose a dynamic spatial-temporal graph attention graph convolutional network (GAGCN) method to improve the generality of the model. The Laplacian matrix in this model is constructed directly by the dependencies among the nodes hidden in the traffic data which are identified by the graph attention networks, and can be dynamic adjust over time, the information of spatial distance among nodes and human intervention are not required in the process. Experimental results of two real-world datasets show that both the generality and prediction accuracy of the proposed model had been significantly improved.