Table 2 Sex classification models using standard cranial measurements.
Classifier and best tuning parameters | Accuracy | Sensitivity | Specificity | PPV | NPV | Selected Features |
|---|---|---|---|---|---|---|
LR {scaled, ‘C’: 1, ‘penalty’: ‘l1’, ‘solver’: ‘liblinear’} | ||||||
CV | 0.90 | 0.89 | 0.90 | 0.90 | 0.89 | RFE, n (features) = 11 GOL, NOL, XCB, ZYB, BBH, BNL, AUB, WFB, NLH_L, MDH_L, ZOB |
Test | 0.91 | 0.95 | 0.88 | 0.88 | 0.95 | |
Test ST | 0.93 | 1.00 | 0.85 | 0.87 | 1.00 | |
Test ZG | 0.90 | 0.90 | 0.90 | 0.90 | 0.90 | |
LDA {scaled, ‘solver’: ‘svd’} | ||||||
CV | 0.91 | 0.90 | 0.93 | 0.93 | 0.91 | RFE, n (features) = 11 GOL, NOL, XCB, ZYB, BNL, AUB, WFB, NLH_L, NLH_R, EKB, MDH_L |
Test | 0.93 | 0.95 | 0.90 | 0.90 | 0.93 | |
Test ST | 0.90 | 1.00 | 0.80 | 0.83 | 0.90 | |
Test ZG | 0.95 | 0.90 | 1.00 | 1.00 | 0.95 | |
SVM {scaled, ‘C’: 1, ‘kernel’: ‘linear’} | ||||||
CV | 0.91 | 0.89 | 0.93 | 0.93 | 0.90 | RFE, n (features) = 18 GOL, NOL, XCB, ZYB, BBH, AUB, WFB, NLH_L, NLH_R, OBB_L, OBH_R, FRC, PAC, OCC, FOL, FOB, MDH_L, ZOB |
Test | 0.91 | 0.93 | 0.90 | 0.90 | 0.92 | |
Test ST | 0.93 | 0.95 | 0.90 | 0.90 | 0.95 | |
Test ZG | 0.90 | 0.90 | 0.90 | 0.90 | 0.90 | |
LR excluded region-specific variables {scaled, ‘C’: 1} | ||||||
CV | 0.87 | 0.87 | 0.87 | 0.87 | 0.87 | RFE, n (features) = 3 ZYB, NLH_L, OBH_R |
Test | 0.90 | 0.90 | 0.90 | 0.90 | 0.90 | |
Test ST | 0.90 | 0.90 | 0.90 | 0.90 | 0.90 | |
Test ZG | 0.90 | 0.90 | 0.90 | 0.90 | 0.90 | |
