Table 1.
Overall and individual class classification accuracies of machine-learning classifiers with effective wavelengths. a
| Classifier | Parameter | Calibration set accuracy (%) | Prediction set accuracy (%) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Healthy | 2 DPI | 4 DPI | 6 DPI | Overall | Healthy | 2 DPI | 4 DPI | 6 DPI | Overall | ||
| SPA-PLS-DA | 6 | 95.00 | 95.00 | 50.00 | 75.00 | 84.17 | 80.00 | 90.00 | 50.00 | 70.00 | 75.00 |
| SPA-RF | 71 | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 | 80.00 | 90.00 | 90.00 | 90.00 | 85.00 |
| SPA-SVM | (0.01, 48.50) | 100.00 | 100.00 | 95.00 | 95.00 | 98.33 | 86.67 | 90.00 | 80.00 | 90.00 | 86.67 |
| SPA-LS-SVM | (33.87, 2.39) | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 | 90.00 | 80.00 | 90.00 | 90.00 | 88.30 |
| SPA-ELM | 44 | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 | 96.67 | 100.00 | 100.00 | 100.00 | 98.33 |
| SPA-BPNN | 5 | 100.00 | 100.00 | 100.00 | 100.00 | 100.00 | 93.33 | 100.00 | 100.00 | 100.00 | 96.67 |
aParameter: number of LVs for partial least squares-discrimination analysis (PLS-DA), number of forest trees for random forest (RF), (C, g) for support vector machine (SVM), (γ, σ2) for least squares support vector machine (LS-SVM), number of nods for extreme learning machine (ELM), and number of neurons in the hidden layer for back propagation neural network (BPNN).