*2.3. Data Analysis*

In this study, the baseline and Savitzky Golay (SG) were used for the preprocessing, respectively. Spectral preprocessing was performed using Unscrambler X10.1 software. The IRIV and SPA were used to extract characteristic wavelengths, and the LS-SVM was adopted to build the model. LS-SVM [33,34] follows the principle of structural risk minimization and transforms the convex quadratic programming problem of traditional support vector machines into the problem of solving a system of linear equations, which reduces the computational complexity. Variable extraction of IRIV and SPA and modeling of LS-SVM were carried out using MATLAB R2020a. The determination coefficient (Rc2) and the root mean square error (RMSEC) of the calibration set, the determination coefficient (Rp2) and the root mean square error (RMSEP) of the prediction set, and the residual predictive deviation (RPD) were used to evaluate the model performance.
