Table 1.
Comparison of predictive performance.
| Models | MAE | RMSE | MAPE % |
|---|---|---|---|
| (1d/2d/3d) | (1d/2d/3d) | (1d/2d/3d) | |
| LSTM [29] | 4.87/4.71/4.63 | 7.21/6.88/6.68 | 9.81/8.80/8.06 |
| GRU [30] | 4.76/4.70/4.61 | 7.14/6.85/6.62 | 9.73/8.69/7.98 |
| GCN [22] | 4.62/4.55/4.43 | 6.86/6.66/6.46 | 9.38/8.51/7.76 |
| STDN [35] | 4.53/4.50/4.42 | 6.84/6.67/6.49 | 9.21/8.47/7.75 |
| T-GCN [38] | 4.51/4.54/4.32 | 6.80/6.77/6.40 | 9.21/8.51/7.58 |
| A3TGCN [39] | 4.48/4.47/4.42 | 6.65/6.57/6.44 | 8.49/7.95/7.77 |
| ASTGCN [42] | 3.72/3.68/3.66 | 5.41/5.36/5.34 | 7.61/7.62/7.46 |
| STGGN [41] | 2.75/2.75/2.73 | 3.57/3.57/3.54 | 4.52/4.41/4.28 |
| DeepMeta-TRI | 2.68/2.61/2.62 | 3.44/3.35/3.39 | 4.37/4.16/4.09 |