Improved partial least squares regression for rapid determination of reducing sugar of potato flours by near infrared spectroscopy and variable selection method

The feasibility of near infrared (NIR) spectroscopy coupled with variable selection methods for rapid determination of reducing sugar content in potato flours was investigated. Monte Carlo uninformative variable elimination (MCUVE), successive projections algorithm (SPA) and genetic algorithm (GA) were performed comparatively to choose characteristic variables associated with reducing sugar distributions. Eighty three and twenty seven samples were used to calibrate models and assess the performance of the models, respectively. Through comparing the performance of partial least squares regression models with new samples, the optimal models of reducing sugar was obtained with 12 selected variables by combination of MCUVE and SPA method. The correlation coefficient of prediction (r(p)) and root mean square errors of prediction (RMSEP) were 0.981 and 0.243, respectively. The results suggest that NIR technique combining with MCUVE and SPA has significant potential to quantitatively analyze reducing sugar in potato flours; moreover, it could indicate the related spectral contributions.