Table 2.
Training and validation accuracy of the best classifier for each scenario.
| Scenario | Labelling method | Class weighting | Model accuracy | |
|---|---|---|---|---|
| Training | Validation | |||
| Scenario A | X | 96.88% | 81.39% | |
| min percentage |  |
96.49% | 86.50% | |
| X | 96.88% | 79.36% | ||
| class availability | |
97.67% | 79.84% | |
| X | 97.49% | 80.62% | ||
| min percentage | |
96.92% | 81.40% | |
| Scenario B | X | 97.69% | 82.17% | |
| class availability | |
97.11% | 79.84% | |
| X | 96.61% | 82.56% | ||
| min percentage | |
96.22% | 78.68% | |
| Scenario A+B | X | 96.71% | 82.56% | |
| class availability | |
96.71% | 82.17% | |
Employment of different labelling methods and use of the class weighting technique are reported.