Table 3 Summary of the results of the cross-validation tests.
From: Assessing the druggability of protein-protein interactions by a supervised machine-learning method
Kernel function | Â | Positives:negatives | |||
|---|---|---|---|---|---|
| Â | Â | All attributes | Top 10 attributes by F-score | ||
| Â | Â | 1:1 | 1:2 | 1:3 | 1:1 |
Linear | Accuracy | 72.05 ± 6.40 | 75.37 ± 4.75 | 79.22 ± 3.78 | 74.91 ± 5.96 |
|  | Sensitivity | 71.54 ± 8.97 | 65.73 ± 7.80 | 60.21 ± 8.48 | 75.34 ± 8.19 |
|  | Specificity | 72.56 ± 8.31 | 80.19 ± 4.96 | 85.56 ± 3.96 | 74.47 ± 8.14 |
Polynomial | Accuracy | 70.86 ± 8.83 | 76.18 ± 6.06 | 81.18 ± 3.98 | 71.74 ± 7.73 |
|  | Sensitivity | 79.85 ± 9.15 | 53.35 ± 28.74 | 52.38 ± 25.58 | 83.29 ± 10.58 |
|  | Specificity | 61.87 ± 18.47 | 87.60 ± 8.01 | 90.78 ± 5.49 | 60.19 ± 20.23 |
Radial basis function | Accuracy | 80.50 ± 4.33 | 83.43 ± 3.22 | 86.37 ± 2.36 | 81.53 ± 4.36 |
|  | Sensitivity | 81.61 ± 5.84 | 65.18 ± 9.37 | 58.67 ± 10.09 | 82.76 ± 6.09 |
|  | Specificity | 79.40 ± 6.64 | 92.55 ± 3.61 | 95.61 ± 2.46 | 80.29 ± 6.51 |
Sigmoid | Accuracy | 63.79 ± 10.87 | 69.68 ± 7.73 | 73.30 ± 6.97 | 63.32 ± 14.62 |
|  | Sensitivity | 62.62 ± 16.32 | 31.69 ± 23.08 | 23.51 ± 19.63 | 61.37 ± 18.06 |
|  | Specificity | 64.96 ± 16.95 | 88.67 ± 10.62 | 89.90 ± 8.93 | 65.27 ± 17.23 |