Table 6.
Overall performance of learning algorithms (kappa) and by topic (AUC).
| Kappa | AD | PD | PE | PC | AE | DI | HS | PR | |
|---|---|---|---|---|---|---|---|---|---|
| NB | 0.50 | 0.83 | 0.99 | 0.95 | 0.94 | 0.79 | 0.80 | 0.64 | 0.73 |
| Rules | 0.47 | 0.8 | 0.89 | 0.86 | 0.73 | 0.70 | 0.74 | 0.48 | 0.61 |
| Tree | 0.50 | 0.8 | 0.97 | 0.90 | 0.88 | 0.72 | 0.78 | 0.61 | 0.68 |
| NB+ | 0.51 | 0.81 | 0.99 | 0.91 | 0.85 | 0.71 | 0.69 | 0.60 | 0.70 |
| Rules+ | 0.49 | 0.81 | 0.97 | 0.91 | 0.68 | 0.70 | 0.68 | 0.74 | 0.57 |
| Tree+ | 0.50 | 0.81 | 0.98 | 0.91 | 0.75 | 0.70 | 0.65 | 0.59 | 0.60 |
| BN | 0.52 | 0.86 | 0.99 | 0.96 | 0.91 | 0.78 | 0.83 | 0.58 | 0.75 |
| SVM | 0.54 | 0.76 | 0.99 | 0.92 | 0.92 | 0.79 | 0.71 | 0.72 | 0.71 |
| Stacking | 0.56 | 0.85 | 0.99 | 0.95 | 0.93 | 0.78 | 0.79 | 0.73 | 0.75 |
Legend: NB = Naïve Bayes; NB+ = Boosted Naïve Bayes; Rules+ = Boosted rules; Tree+ = Boosted decision tree; BN = Bayesian network; SVM = Support Vector Machine; AD = adult dose; PD = pediatric dose; PE = patient education; PC = pregnancy category; AE = adverse effects; DI = drug interactions; HS = how supplied; PR = precautions.