Towards Spatio-Temporal Cross-Platform Graph Embedding Fusion for Urban Traffic Flow Prediction

To address the above challenges, we propose STC-GEF, the novel Spatio-Temporal Cross-platform Graph Embedding Fusion approach In this paper, we have proposed STC-GEF, a novel Spatio-Temporal for the urban taxi flow prediction. In this prototype study, Cross-platform Graph Embedding Fusion approach for the urban we have made the following three major contributions: traffic flow prediction. We have designed a spatial embedding module based on graph convolutional networks (GCN) to extract the (1) Spatial and Temporal Graph Embedding Learning: To complex spatial features within the traffic flow data. Furthermore, extract the complex spatial features within traffic flow data, to capture the temporal dependencies between the traffic flow data we propose a spatial embedding module based on graph from various time intervals, we have designed a temporal embedding convolutional networks (GCN). Additionally, to capture the module based on recurrent neural networks. Based on the temporal dependencies between the traffic flow data from observations that different transportation platforms' trip data (e.g., various time intervals, we leverage a temporal embedding taxis, Uber, and Lyft) can be correlated, we have designed an effective module based on recurrent neural networks.