用于多元时间序列预测的图神经网络模型
|
|
张晗
|
Graph neural network model for multivariate time series forecasting
|
|
Han ZHANG
|
|
| 表 2 多步预测任务中所有模型的实验结果 |
| Tab.2 Experimental results of models for multi-step forecasting |
|
| 模型 | MAE | | MAPE/% | | RMSE | | METR-LA | PEMS-BAY | METR-LA | PEMS-BAY | METR-LA | PEMS-BAY | | VAR | 3.60 | 2.07 | | 4.90 | 10.50 | | 4.74 | 27.60 | | DSANet | 4.59 | 2.49 | 4.90 | 12.70 | 5.69 | 29.40 | | DCRNN | 3.53 | 1.95 | 5.79 | 10.01 | 4.52 | 27.37 | | STGCN | 3.59 | 2.20 | 4.63 | 10.63 | 5.06 | 27.11 | | ASTGCN | 3.49 | 1.91 | 5.45 | 10.01 | 4.46 | 28.07 | | STSGCN | 3.40 | 1.95 | 4.60 | 10.05 | 4.49 | 26.88 | | AGCRN | 3.49 | 1.94 | 4.53 | 9.87 | 4.47 | 25.24 | | MTSGNN-S | 3.10 | 1.89 | 4.38 | 9.25 | 4.37 | 21.85 | | MTSGNN-D | 3.15 | 1.91 | 4.45 | 9.25 | 4.42 | 22.12 | | MTSGNN-GL | 3.09 | 1.88 | 3.37 | 9.17 | 3.38 | 21.81 | | MTSGNN | 3.05 | 1.75 | 4.01 | 8.72 | 4.02 | 20.72 |
|
|
|