Local linear embedded regression in the quantitative analysis of glucose in near infrared spectra

This paper investigates the use of Local Linear Embedded Regression (LLER) for the quantitative analysis of glucose from near infrared spectra. The performance of the LLER model is evaluated and compared with the regression techniques Principal Component Regression (PCR), Partial Least Squares Regression (PLSR) and Support Vector Regression (SVR) both with and without pre-processing. The prediction capability of the proposed model has been validated to predict the glucose concentration in an aqueous solution composed of three components (urea, triacetin and glucose). The results show that the LLER method offers improvements in comparison to PCR, PLSR and SVR. This paper investigates the use of local linear embedded regression in the context of determination of glucose from NIR spectra.