Quantitative Analysis of Fat in Milk by UV-Vis-NIR Reflectance Spectroscopy and Multivariate Analysis

Introduction

Milk is one of the most popular drinks among humans. In recent years, however, there has been a great increase in the sales of milk products with adjusted fat content. Milk fat content is typically measured using the Roese-Gottlieb or the Gerber method, but measurement by these methods is extremely time-consuming. We therefore investigated the use of the spectral reflectance method as a simpler quantitative method. By applying a combination of reflectance measurement using a screw-top glass tube in conjunction with multivariate analysis, we found that the fat content could be determined quite easily. As multivariate analysis methods, the multiple linear regression method, PLS method, and support vector regression method (SVR) were used to conduct a comparative analysis of the quantitative accuracy of these quantitative methods. The results indicated that the support vector regression method provided the best quantitative accuracy. The results are introduced in this paper.

November 15, 2013 GMT