DDP-GCN: Multi-graph convolutional network for spatiotemporal traffic forecasting

Traffic speed forecasting is one of the core problems in transportation systems. For a more accurate prediction, recent studies started using not only temporal patterns but also spatial information on road network through graph convolutional networks. Even though highly complex due to its non-Euclidean and directional characteristics, previous approaches mainly focused modeling dependencies distance only. In this paper, we identify two essential traffic addition distance, direction positional relationship, for designing basic elements as fundamental building blocks. Using blocks, suggest DDP-GCN (Distance, Direction, Positional relationship Graph Convolutional Network) incorporate three relationships into deep neural We evaluate proposed model with large-scale real-world datasets, find positive improvements long-term urban The improvement can be larger commute hours, it limited short-term forecasting.