Table 2.
Results of the PLSR model based on different preprocessing methods.
| Preprocessing Methods |
The Number of Optimal Principal Components | Rc | RMSEc (g/kg) | Rp | RMSEp (g/kg) | RPD |
|---|---|---|---|---|---|---|
| X | 11 | 0.90 | 0.3 | 0.90 | 0.3 | 2.3 |
| SG | 11 | 0.90 | 0.3 | 0.90 | 0.3 | 2.3 |
| MSC | 10 | 0.91 | 0.3 | 0.91 | 0.3 | 2.3 |
| SNV | 10 | 0.91 | 0.3 | 0.91 | 0.3 | 2.3 |
| 1D | 8 | 0.92 | 0.3 | 0.90 | 0.3 | 2.3 |
| 1D + SG | 8 | 0.90 | 0.3 | 0.90 | 0.3 | 2.3 |
| 1D + MSC | 8 | 0.94 | 0.24 | 0.92 | 0.3 | 3 |
| 1D + SNV | 8 | 0.94 | 0.24 | 0.92 | 0.3 | 3 |
Note: Savitzky−Golay (S−G) convolution smoothing, multiplicative scatter correction (MSC), standard normal variate transformation (SNV), the first derivative(1D), the first derivative combined with S-G smoothing (1D + SG), the first derivative combined with multiplicative scatter correction (1D + MSC), and the first derivative combined with standard normal variate transformation (1D + SNV).