Table 2. Performance of PLS, SVM and BPNN models for predicting firmness and TSS.
Calibration | Prediction | |||||
Parameter | Method | Spectra | SEC | r | SEP | r |
Firmness(N) | PLS | VIS | 0.518 | 0.91 | 0.451 | 0.93 |
NIR | 0.859 | 0.81 | 0.943 | 0.67 | ||
Whole | 0.504 | 0.91 | 0.421 | 0.94 | ||
SVM | VIS | 0.149 | 0.94 | 0.674 | 0.76 | |
NIR | 1.523 | 0.78 | 0.794 | 0.72 | ||
Whole | 0.112 | 0.94 | 0.532 | 0.78 | ||
BPNN | VIS | 0.599 | 0.90 | 0.436 | 0.93 | |
NIR | 0.677 | 0.87 | 0.713 | 0.87 | ||
Whole | 0.424 | 0.93 | 0.375 | 0.94 | ||
TSS (°Bx) | PLS | VIS | 0.776 | 0.64 | 0.638 | 0.73 |
NIR | 0.783 | 0.55 | 0.795 | 0.67 | ||
Whole | 0.712 | 0.81 | 0.575 | 0.76 | ||
SVM | VIS | 0.756 | 0.67 | 0.784 | 0.60 | |
NIR | 2.780 | 0.48 | 0.855 | 0.42 | ||
Whole | 0.361 | 0.92 | 0.775 | 0.62 | ||
BPNN | VIS | 0.721 | 0.59 | 0.656 | 0.78 | |
NIR | 0.753 | 0.58 | 0.705 | 0.74 | ||
Whole | 0.506 | 0.76 | 0.573 | 0.83 |