Skip to main content

Table 6 Comparison between SDA-LM, SAE-LM and SDA structures on imbalanced dataset

From: Flight delay prediction based on deep learning and Levenberg-Marquart algorithm

 

Accuracy

Precision

Recall

Specificity

F1 measure

Support

SDA-LM model

 Class 0

0.92

0.98

0.88

0.89

0.92

2888640

 Class1

0.92

0.74

0.89

0.88

0.84

713039

 Micavg

0.92

0.88

0.88

0.89

0.88

3601679

 Macavg

0.92

0.89

0.88

0.88

0.83

3601679

 Weighted Avg

0.92

0.90

0.88

0.88

0.89

3601679

SAE-LM model

 Class 0

0.83

0.92

0.86

0.68

0.89

2888640

 Class1

0.83

0.55

0.68

0.86

0.61

713039

 Micavg

0.83

0.83

0.83

0.74

0.83

3601679

 Macavg

0.83

0.73

0.77

0.77

0.75

3601679

 Weighted Avg

0.83

0.84

0.83

0.75

0.83

3601679

SAE-LM model

 Class 0

0.80

0.93

0.82

0.73

0.87

2888640

 Class1

0.80

0.50

0.74

0.82

0.60

713039

 Micavg

0.80

0.80

0.80

0.78

0.80

3601679

 Macavg

0.80

0.72

0.78

0.77

0.73

3601679

 Weighted Avg

0.80

0.84

0.80

0.78

0.82

3601679