Application of two near-infrared instruments combined with chemometrics and key variable selection algorithms for the determination of anthocyanins content in black rice

Anthocyanins in black rice are important for human health, and the anthocyanin content determines the nutritional quality of black rice. In this study, two NIR spectra (Spectrometer A, interval of 1 nm, 680-2600 nm; Spectrometer B, interval of 4 nm, 400-2496 nm) combined with principal component regression (PCR), partial least squares regression (PLSR), radial basis function (RBF) and linear kernel function of support vector machine regression (SVR) were used to predict anthocyanin content in black rice. Comparing the predictive accuracy of the combination of different instruments and chemometrics for anthocyanin content can help researchers achieve rapid and accurate determinations. Our results indicated that the prediction accuracy of Spectrometer B is higher. In the full NIR spectrum, the accuracy of the standard normal variate (SNV) pretreatment combined with linear-SVR reached 0.972. Three algorithms were applied to select critical variables, including random frog (RF), competitive adaptive reweighted sampling (CARS), and Monte Carlo uninformative variable elimination (MCUVE). The RF algorithm screened 10 key variables and the PCR model based on the key variables had the highest accuracy of 0.934. Our study provided a theoretical basis for the rapid determination of anthocyanin content of black rice.