Research Article
PCG Classification Using Multidomain Features and SVM Classifier
Table 13
Performance of the classification.
| Case | Percent of data to train | Percent of data to test | Repeat times | Training and test data division | Sensitivity | Specificity | Overall score |
| Case 1 | 100% | 100% | 1 | No | 0.99 | 0.91 | 0.95 |
| Case 2 | 10% | 90% | 200 | Yes | 0.68±0.06 | 0.87±0.03 | 0.77±0.02 | 20% | 80% | 200 | Yes | 0.76±0.05 | 0.86±0.02 | 0.81±0.02 | 30% | 70% | 200 | Yes | 0.80±0.04 | 0.87±0.02 | 0.83±0.02 | 40% | 60% | 200 | Yes | 0.82±0.04 | 0.87±0.01 | 0.85±0.02 | 50% | 50% | 200 | Yes | 0.84±0.03 | 0.87±0.01 | 0.85±0.01 | 60% | 40% | 200 | Yes | 0.85±0.04 | 0.87±0.01 | 0.86±0.01 | 70% | 30% | 200 | Yes | 0.86±0.04 | 0.87±0.01 | 0.87±0.02 | 80% | 20% | 200 | Yes | 0.87±0.04 | 0.87±0.02 | 0.87±0.02 | 90% | 10% | 200 | Yes | 0.88±0.04 | 0.87±0.02 | 0.88±0.02 |
|
|
Note: the number is presented as mean±SD.
|