Table 3 The performance of the prediction models based on different classifications on the independent tests for two ratios
Train/test ratios | Algorithms | Sensitivity | Specificity | Precision | F1 | AUC (95% CI) | Accuracy (95% CI) |
|---|---|---|---|---|---|---|---|
80/20 | GLM | 0.585 | 0.169 | 0.399 | 0.475 | 0.630 (0.619, 0.641) | 0.371 (0.359, 0.383) |
Ridge | 0.503 | 0.789 | 0.683 | 0.580 | 0.699 (0.686, 0.713) | 0.645 (0.633, 0.658) | |
Lasso | 0.484 | 0.814 | 0.711 | 0.576 | 0.711 (0.698, 0.724) | 0.654 (0.641, 0.666) | |
elastic-net | 0.484 | 0.802 | 0.697 | 0.572 | 0.701 (0.689, 0.714) | 0.647 (0.635, 0.660) | |
NN | 0.499 | 0.785 | 0.686 | 578 | 0.697 (0.684, 0.711) | 0.646 (0.634, 0.658) | |
RF | 0.524 | 0.819 | 0.732 | 0.611 | 0.756 (0.744, 0.769) | 0.676 (0.663, 0.688) | |
70/30 | GLM | 0.601 | 0.189 | 0.361 | 0.445 | 0.653 (0.639, 0.667) | 0.356 (0.344, 0.369) |
Ridge | 0.510 | 0.804 | 0.743 | 0.604 | 0.703 (0.690, 0.717) | 0.649 (0.636, 0.661) | |
Lasso | 0.516 | 0.819 | 0.698 | 0.593 | 0.717 (0.704, 0.730) | 0.683 (0.671, 0.695) | |
Elastic-net | 0.527 | 0.824 | 0.717 | 0.608 | 0.720 (0.707, 0.733) | 0.682 (0.670, 0.694) | |
NN | 0.499 | 0.785 | 0.751 | 0.621 | 0.701 (0.688, 0.715) | 0.656 (0.644, 0.668) | |
RF | 0.524 | 0.819 | 0.715 | 0.595 | 0.761 (0.749, 0.773) | 0.688 (0.676, 0.700) |