Table 8. The testing performance of the CNN_4 on the GWH dataset with different window sizes.
The testing results of the CNN_4 performed on the hold-out data with different window sizes.
| Site | Window size (nt) | Model | |||
|---|---|---|---|---|---|
| CNN_4 | |||||
| AUPRC | Precision | Recall | MCC | ||
| Donor | 40 | 0.922 | 0.846 | 0.882 | 0.844 |
| 80 | 0.931 | 0.868 | 0.886 | 0.859 | |
| 160 | 0.945 | 0.884 | 0.901 | 0.877 | |
| 240 | 0.950 | 0.887 | 0.910 | 0.883 | |
| 398 | 0.950 | 0.887 | 0.912 | 0.885 | |
| Acceptor | 40 | 0.840 | 0.762 | 0.786 | 0.741 |
| 80 | 0.887 | 0.813 | 0.831 | 0.796 | |
| 160 | 0.919 | 0.854 | 0.858 | 0.834 | |
| 240 | 0.932 | 0.878 | 0.867 | 0.855 | |
| 398 | 0.937 | 0.883 | 0.878 | 0.863 | |
Notes.
Bold styling emphasizes the highest values regarding the evaluation metrics used in the study.