Table 1 Performance comparison across different prediction horizons.
From: MeshHSTGT: hierarchical spatio-temporal fusion for mesh network traffic forecasting
Model | Mesh Data (10 min/30 min/60 min) | ||
|---|---|---|---|
MAE | MAPE | RMSE | |
FNN | 154.98/387.25/813.28 | 0.24/0.25/0.30 | 296.83/472.83/1025.48 |
FC-LSTM | 172.57/387.25/781.52 | 0.21/0.20/0.27 | 248.14/457.19/954.39 |
GWN | 114.38/220.25/513.93 | 0.16/0.17/0.18 | 144.58/326.05/625.11 |
STChebNet | 123.49/251.25/584.94 | 0.17/0.20/0.21 | 181.34/331.64/647.62 |
STSGCN | 101.14/224.12/529.35 | 0.14/0.18/0.19 | 139.05/274.58/561.79 |
STFGNN | 103.56/212.83/537.81 | 0.14/0.17/0.19 | 146.67/261.31/542.34 |
TSGAN | 94.71/201.08/420.74 | 0.11/0.12/0.14 | 121.41/233.72/493.56 |
Ours | 89.62/193.41/391.17 | 0.09/0.10/0.10 | 113.24/235.09/470.25 |
