Table 1 Comparison of different methods on the SHMetro dataset.
From: TSTA-GCN: trend spatio-temporal traffic flow prediction using adaptive graph convolution network
Time | 15 min | 30 min | 60 min | ||||||
|---|---|---|---|---|---|---|---|---|---|
Metric | RMSE | MAE | MAPE | RMSE | MAE | MAPE | RMSE | MAE | MAPE |
HA | 136.97 | 48.26 | 31.55% | 136.81 | 47.88 | 31.49% | 135.72 | 46.40 | 30.80% |
RF | 66.63 | 34.37 | 24.09% | 88.03 | 41.37 | 28.89% | 143.5 | 59.15 | 52.91% |
MLP | 48.71 | 25.16 | 19.44% | 51.80 | 26.15 | 20.38% | 63.33 | 29.92 | 23.96% |
LSTM | 55.53 | 26.68 | 18.76% | 57.37 | 27.25 | 19.04% | 63.41 | 28.94 | 20.59% |
GRU | 52.04 | 25.91 | 18.87% | 54.20 | 26.39 | 19.20% | 59.91 | 28.08 | 21.03% |
ASTGCN | 66.49 | 32.29 | 21.90% | 98.76 | 39.28 | 25.63% | 154.95 | 51.33 | 32.35% |
STG2Seq | 47.19 | 24.98 | 23.26% | 50.58 | 26.17 | 26.79% | 56.81 | 28.22 | 34.30% |
DCRNN | 46.02 | 24.04 | 17.82% | 49.90 | 25.23 | 18.35% | 58.83 | 28.01 | 20.44% |
GCRNN | 46.09 | 24.26 | 18.06% | 50.12 | 25.42 | 18.73% | 58.67 | 28.18 | 21.07% |
GWN | 46.98 | 24.91 | 20.05% | 51.64 | 26.53 | 20.38% | 65.08 | 30.90 | 24.36% |
PVCGN | 44.97 | 23.29 | 16.83% | 47.83 | 24.16 | 17.23% | 55.27 | 26.29 | 18.69% |
MGT | 45.30 | 23.15 | 16.47% | 46.80 | 23.45 | 16.53% | 50.69 | 24.97 | 17.83% |
ASC-GRU | 51.68 | 25.13 | 18.66% | 52.12 | 26.29 | 19.01% | 56.02 | 27.86 | 20.76% |
TSTA-GCN | 42.58 | 22.63 | 17.24% | 44.99 | 23.24 | 17.06% | 48.54 | 24.52 | 17.97% |
