Table 3 Performance comparison of different models.

From: Traffic flow prediction based on temporal attention and multi-graph adjacency fusion using DynamicChebNet

Baseline

PeMS04

PeMS08

MAE

MAPE (%)

RMSE

MAE

MAPE/%

RMSE

DCRNN

22.75

14.84

36.83

18.30

11.35

28.52

STGCN

21.82

13.91

34.91

17.87

11.21

27.37

T-GCN

20.95

13.80

33.37

16.55

11.08

25.73

GWNET

19.45

13.44

31.75

15.53

10.61

24.85

GMAN

19.16

13.21

31.62

15.44

10.03

24.59

MTGNN

19.08

12.95

31.46

15.38

10.17

24.93

LightCTS

18.81

12.82

31.21

15.13

9.85

24.43

ASTGNN

18.65

12.63

31.04

14.98

9.62

24.89

DDGCRN

18.59

12.51

30.64

14.67

9.45

23.75

STID

18.55

12.49

29.83

14.22

9.42

23.53

PDFormer

18.52

12.46

29.97

13.85

9.41

23.22

TMDCN

18.33

12.25

29.23

13.72

9.37

22.81

Back to article page