Table 3.
Predictive performance of the selected binary logistic regression model for detecting woody breast (WB) condition in broiler carcasses by strain using image measurements.
| Strain | Data set1 | Fit details2 |
Parameter estimates and odds ratio (OR) |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Gen. R2 | RMSE | MAD | MR (%) | TPR (%) | FPR (%) | AUC | Parameter | Estimate | Std. Error | Pr > χ2 | OR | ||
| High breast-yielding (HBY) | Training | 0.65 | 0.28 | 0.15 | 9.76 | 68.54 | 3.93 | 0.95 | M1 | 1.53 | 0.35 | <0.0001 | 4.64 |
| Validation | 0.62 | 0.27 | 0.15 | 8.89 | 73.53 | 4.79 | 0.93 | M2 | −8.03 | 1.19 | <0.0001 | 3.24 × 10−4 | |
| M3 | 1.84 | 0.31 | <0.0001 | 6.30 | |||||||||
| Standard breast-yielding (SBY) | Training | 0.59 | 0.21 | 0.09 | 5.71 | 60.00 | 2.11 | 0.96 | M1 | 1.37 | 0.62 | 0.0277 | 3.95 |
| Validation | 0.65 | 0.19 | 0.08 | 5.56 | 62.50 | 2.44 | 0.98 | M2 | −7.95 | 2.12 | 0.0002 | 3.53 × 10−4 | |
| M3 | 1.84 | 0.50 | 0.0003 | 6.31 | |||||||||
Abbreviations: AUC, area under the ROC (receiver operating characteristic) curve; FPR, false positive rate; MAD, mean absolute deviation; MR, misclassification rate; RMSE, root mean square error; TPR, true positive rate (sensitivity, ST).
1
Training (HBY: n = 420; SBY: n = 210) and validation (HBY: n = 180; SBY: n = 90).
2
The model used for WB prediction (Model 1): Logit (p) = α + β1 M1 + β2 M2 + β3 M3.