Table 3 DNN classification results of each resampling strategy with different modulation factor
Resampling strategy | Modulation factor γ | Accuracy | Precision | Recall | F1-score |
|---|---|---|---|---|---|
Raw | 89.12 | 85.33 | 75.34 | 78.63 | |
SMOTE | 0.6 | 91.56 | 85.58 | 85.45 | 85.44 |
0.85 | 92.03 | 88.36 | 84.23 | 85.69 | |
1 | 92.59 | 88.40 | 85.07 | 86.60 | |
ADASYN | 0.6 | 92.03 | 84.64 | 85.23 | 84.88 |
0.85 | 93.06 | 87.70 | 85.27 | 86.41 | |
1 | 92.31 | 86.18 | 84.85 | 85.50 | |
BorderlineSMOTE | 0.6 | 92.12 | 85.97 | 82.74 | 84.12 |
0.85 | 93.15 | 88.46 | 85.68 | 86.98 | |
1 | 93.06 | 88.36 | 84.06 | 85.91 | |
SMOTEENN | 0.6 | 88.84 | 80.17 | 83.34 | 81.64 |
0.85 | 89.96 | 81.93 | 84.17 | 82.96 | |
1 | 89.68 | 83.77 | 84.93 | 84.09 | |
SMOTETomek | 0.6 | 92.31 | 85.92 | 85.18 | 85.43 |
0.85 | 92.87 | 87.80 | 85.93 | 86.80 | |
1 | 92.78 | 86.54 | 84.86 | 85.61 |