Fig 1.
Spatio-temporal network prediction task.
Fig 2.
Efficient Channel Attention Mechanism.
Fig 3.
Framework of Transformer-based Hypergraph Convolutional Traffic Flow Prediction Model(The model architecture employs three stacked S-T Blocks).
Fig 4.
Constructing a hypergraph.
Table 1.
Table caption Nulla mi mi, venenatis sed ipsum varius, volutpat euismod diam.
Table 2.
Information of PeMSD4 dataset and PeMSD8 dataset.
Table 3.
Performance of different models on the PeMSD4 and PeMSD8 datasets. Results are presented as mean ± standard deviation (5 independent runs).
Table 4.
Paired t-test results between TSHGCN and the strongest baseline model.
Fig 5.
Hypergraph visualization.
Fig 6.
Sensitivity Analysis of Hypergraph Neighborhood Size Parameter k.
Fig 7.
Sensitivity Analysis of Hypergraph Convolutional Layer Count L.
Table 5.
Ablation Study Results on PeMSD4 and PeMSD8 Datasets.
Fig 8.
Heatmap Analysis of Joint Effects of Hypergraph Parameters k and L.
Fig 9.
Hyperedge Weight Spatiotemporal Dynamic Evolution.
Fig 10.
Ablation experiment on PeMSD4 and PeMSD8 datasets.
Table 6.
Ablation Study Results on PeMSD4 and PeMSD8 Datasets.