Table 1.
Comparison of predicting performance on the testing set using Logistic Regression.
| Spectral region (cm−1) | Spectra signal | Accuracy (%) | Sensitivity (%) | Specificity (%) | AUC (× 100) | Accuracy (%) | Sensitivity (%) | Specificity (%) | AUC (× 100) |
|---|---|---|---|---|---|---|---|---|---|
| Out-of-sample | In-sample | ||||||||
| 600 to 4500 | Raw | 95.7 | 94.1 | 96.7 | 98.4 | 100 | 100 | 100 | 100 |
| Baseline Corrected | 96.8 | 94.1 | 98.3 | 99.8 | 100 | 100 | 100 | 100 | |
| 1st Der | 96.8 | 97 | 98.3 | 99.9 | 100 | 100 | 100 | 100 | |
| 2nd Der | 97.8 | 97 | 98.3 | 99.1 | 100 | 100 | 100 | 100 | |
| 900 to 1800 | Raw | 83.1 | 70.5 | 86.5 | 89 | 92.1 | 82.3 | 100 | 94.3 |
| Baseline Corrected | 84.2 | 73.5 | 86.5 | 90.5 | 94.7 | 88.2 | 100 | 95.5 | |
| 1st Der | 93.6 | 94.1 | 92.4 | 97.7 | 100 | 100 | 100 | 100 | |
| 2nd Der | 94.7 | 94.1 | 95.4 | 97.2 | 100 | 100 | 100 | 100 | |