Table 3 Traffic flow prediction results of different methods on PEMS03 and PEMS07.

From: Temporal representation learning enhanced dynamic adversarial graph convolutional network for traffic flow prediction

Methods

PEMS03

PEMS07

MAE

RMSE

MAPE (%)

MAE

RMSE

MAPE (%)

VAR

23.65

38.26

24.51

50.22

75.63

32.22

SVR

21.97

35.29

21.51

32.49

45.84

13.20

ARIMA

35.41

47.59

33.78

38.17

59.27

19.46

LSTM

20.62

33.54

28.94

29.71

45.32

14.14

GCRN

19.88

33.20

19.71

31.03

48.70

15.67

STGCN

17.55

30.42

17.34

25.33

39.34

11.21

DCRNN

17.99

30.31

18.34

25.22

38.61

11.82

ASTGCN(r)

17.34

29.56

17.21

24.01

37.87

10.73

OGCRNN

17.16

29.40

16.48

24.45

39.80

10.22

STSGCN

17.48

29.21

16.78

24.26

39.03

10.21

GWN

19.12

32.77

18.89

26.39

41.50

11.97

Z-GCNETs

16.64

28.15

16.39

21.77

35.17

9.25

AGCRN

16.03

28.52

14.65

22.37

35.70

9.55

TRL-DAG (ours)

14.78

25.90

14.13

20.83

34.19

8.65

Back to article page