A Spatiotemporal Traffic Speed Prediction Method Combining Attention and Multivariate Graph Convolution Fusion

Abstract Spatiotemporal prediction is widely used in the fields of neuroscience, climate, and transportation, traffic speed one typical research areas. Since networks are irregular grid structures with complex nonlinear spatiotemporal dependencies between nodes, traditional single-feature methods difficult to adapt road conditions. In order improve accuracy speed, this paper proposes a method combining attention multivariate graph convolution fusion (CAMGCF). It constructs model, specifically, network uses fuse external factor features an mechanism adaptively feature information from different time series achieve stability long-term prediction. Finally, framework evaluated on three real-world large-scale datasets, experimental results show that has more accurate compared benchmark method.