A new ensemble deep graph reinforcement learning network for spatio-temporal traffic volume forecasting in a freeway network

Pan Shang, Xinwei Liu, Chengqing Yu, Guangxi Yan, Qingqing Xiang, Xiwei Mi

Abstract: Highlights•A spatiotemporal traffic volume forecasting model is proposed in freeway network.•GCN and GAT is used as the main predictor.•DQN is used to integrate GCN and GAT.•The proposed model is compared with twenty-one mainstream forecasting models.