Table 2.
PLS-DA classification model performance for microwave pretreated mustard using SNV + SG smoothened Vis–NIR HSI spectral data and SNV + SGD2 smoothened SWIR HSI spectral data, with 10 and 3 latent variables, respectively.
| Calibration | Cross validation | Testing | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Untreated | MW1 | MW2 | Untreated | MW1 | MW2 | Untreated | MW1 | MW2 | |
| Vis–NIR | |||||||||
| Sensitivity | 0.991 | 0.971 | 0.934 | 0.991 | 0.971 | 0.934 | 0.992 | 0.976 | 0.973 |
| Specificity | 0.995 | 0.964 | 0.990 | 0.995 | 0.964 | 0.989 | 0.998 | 0.984 | 0.989 |
| Precision | 0.991 | 0.936 | 0.976 | 0.991 | 0.936 | 0.976 | 0.996 | 0.972 | 0.974 |
| Accuracy | 0.966 | 0.966 | 0.981 | ||||||
| Non-assigned | 0.116 | 0.116 | 0.087 | ||||||
| SWIR | |||||||||
| Sensitivity | 1.000 | 0.998 | 0.990 | 1.000 | 0.997 | 0.989 | 1.000 | 0.984 | 1.000 |
| Specificity | 1.000 | 0.995 | 0.999 | 1.000 | 0.995 | 0.998 | 1.000 | 1.000 | 0.992 |
| Precision | 1.000 | 0.990 | 0.997 | 1.000 | 0.990 | 0.997 | 1.000 | 1.000 | 0.981 |
| Accuracy | 0.996 | 0.995 | 0.994 | ||||||
| Non-assigned | 0.011 | 0.012 | 0.006 | ||||||