To overcome these limitations, we propose a Spatial-Temporal Interactive Dynamic Graph Convolutional Network (STIDGCN) for traffic forecasting. Specifically, we propose an interactive learning ...