Best performance achieved by each algorithm for classifying patient identification incidents
| Algorithm | Accuracy | Precision | Recall | F-measure | Area under curve | Feature selection methods | Training set size | |||||||
| R | E | R | E | R | E | R | E | R | E | R | E | R | E | |
| Naïve Bayes | 96.77 | 95.97 | 0.95 | 0.95 | 0.98 | 0.98 | 0.97 | 0.96 | 0.99 | 0.99 | 1 | 1, 2 | 500 | 500 |
| Support Vector Machine (linear) | 97.58 | 91.53 |
| 0.97 |
| 0.86 |
| 0.91 |
| 0.97 |
| 1, 2 |
| 200 |
| Support Vector Machine (radial-basis function) | 97.98 | 94.35 | 0.98 | 0.97 | 0.98 | 0.92 | 0.98 | 0.94 | 1.00 | 0.99 | 1, 2, 3 | 1, 2 | 300 | 100 |
E, performance tested against tested against expert-classified handover incidents; R, performance tested against reporter-classified incidents.