Table 3 Model performance metrics.

From: Continuous-time air pollutant forecasting using multi-timescale attention neural ordinary differential equations (MA-NODE)

Pollutant

Step

R²

MSE

RMSE

MAE

Coverage

Calibration Error

O₃ (ppb)

1

0.96

4.08

2.02

1.46

0.91

0.04

 

2

0.96

3.61

1.90

1.45

0.86

0.09

 

3

0.83

15.82

3.98

3.09

0.88

0.07

CO (ppm)

1

0.93

0.02

0.16

0.12

0.93

0.02

 

2

0.91

0.03

0.18

0.13

0.88

0.07

 

3

0.49

0.18

0.42

0.32

0.89

0.06

NO₂ (ppb)

1

0.95

11.33

3.37

2.55

0.93

0.02

 

2

0.93

15.48

3.93

2.80

0.89

0.06

 

3

0.50

103.56

10.18

7.71

0.91

0.04

SO₂ (ppb)

1

0.90

0.45

0.67

0.46

0.93

0.02

 

2

0.84

0.77

0.88

0.61

0.90

0.05

 

3

0.44

2.65

1.63

1.17

0.90

0.05

PM₁₀ (µg/m³)

1

0.95

49.30

7.02

5.35

0.92

0.03

 

2

0.94

59.21

7.70

5.58

0.86

0.09

 

3

0.53

456.39

21.36

16.08

0.90

0.05

PM₂.₅ (µg/m³)

1

0.95

7.38

2.72

1.94

0.93

0.02

 

2

0.93

10.68

3.27

2.24

0.88

0.07

 

3

0.50

80.17

8.95

6.33

0.90

0.05

Back to article page