Method | Performance (%) on Unbalanced training set - 1:10 | |||
---|---|---|---|---|
Accuracy | Precision | Recall | Specificity | |
Decision trees | 96.63 ± 0.46 | 91.66 ± 1.47 | 73.14 ± 2.93 | 99.26 ± 0.14 |
Random forest | 97.85 ± 0.25 | 95.87 ± 0.68 | 82.12 ± 2.28 | 99.61 ± 0.06 |
DeepPPI | 98.09 ± 0.77 | 95.30 ± 3.86 | 83.29 ± 7.49 | 99.58 ± 0.37 |
DeepFE-PPI | 98.50 ± 0.46 | 96.56 ± 0.38 | 87.86 ± 5.00 | 99.66 ± 0.05 |
Struct2Graph | 99.26 ± 0.15 | 97.04 ± 0.70 | 95.59 ± 0.73 | 99.67 ± 0.10 |
Method | MCC | F1-score | ROC-AUC | NPV |
---|---|---|---|---|
Decision trees | 80.13 ± 2.06 | 81.33 ± 1.99 | 86.20 ± 1.48 | 97.07 ± 0.45 |
Random forest | 87.60 ± 1.12 | 88.44 ± 1.11 | 99.49 ± 0.22 | 98.03 ± 0.32 |
DeepPPI | 88.01 ± 4.96 | 88.69 ± 4.82 | 96.65 ± 1.60 | 98.35 ± 0.73 |
DeepFE-PPI | 91.27 ± 2.80 | 91.92 ± 2.72 | 99.50 ± 0.20 | 98.69 ± 0.24 |
Struct2Graph | 95.90 ± 0.60 | 96.31 ± 0.52 | 99.54 ± 0.22 | 99.50 ± 0.12 |