Table 1.
For case 1: Average F1 scores, precisions, and recalls of the risk gene prediction models built on SASMOTES, SASMOTE without visible neighbors, SASMOTE without inspections, B-SMOTE, SMOTE, and original datasets. The values in brackets represent the standard deviation (Std) in 5-fold cross validation. The proposed SASMOTE performs better on the average recall and F1 score, and the original dataset performs best on the average precision
| Model | Precision % (Std) | Recall % (Std) | F1 score % (Std) |
|---|---|---|---|
| SASMOTE | 52.57 (8.19) | 50.05 (5.38) | 51.16 (6.32) |
| SASMOTE w/o invisible | 50.34 (10.10) | 49.37 (10.06) | 49.72 (9.67) |
| SASMOTE w/o inspection | 53.02 (12.30) | 47.92 (11.80) | 50.26 (11.94) |
| B-SMOTE | 49.12 (10.16) | 47.20 (8.76) | 47.80 (8.12) |
| SMOTE | 46.85 (11.89) | 45.47 (10.97) | 46.07 (11.19) |
| Original data | 68.53 (9.31) | 22.64 (6.06) | 33.83 (7.70) |