Table 1 Performances of the 2-level classifiers for the training dataset
| Â | Classifier performance | |||
|---|---|---|---|---|
SVM | KNN | Decision tree | ||
AUC | IC vs nonIC | 0.99 ~ 1 | 0.99 ~ 1 | 0.99 |
IP vs NT | 0.96 ~ 0.98 | 0.98 | 0.78 ~ 0.80 | |
Accuracy | IC vs nonIC | 96.3 ~ 96.7% | 96.4 ~ 96.6% | 95.4 ~ 95.8% |
IP vs NT | 95.0 ~ 95.9% | 94.3 ~ 94.8% | 84.3 ~ 85.5% | |
Sensitivity | IC vs nonIC (true rate for IC) | 95 ~ 96% | 95 ~ 96% | 94 ~ 96% |
IP vs NT (true rate for IP) | 85 ~ 86% | 80 ~ 81% | 30 ~ 36% | |
Specificity | IC vs nonIC (true rate for nonIC) | 97 ~ 98% | 97% | 97% |
IP vs NT (true rate for NT) | 97 ~ 98% | 98% | 95 ~ 97% | |