2009 Volume 26 Issue 5 Pages 451-457
The near-infrared spectroscopy is used to analyze various foods because of the facts that the measurement time is very short and the coverage area is very wide. This analysis technique is recognized as an important technique to finger printing in addition to the element measurement. This research aimed at the development of the data analysis software for metabolic finger printing of food using the near-infrared spectroscopy. This software was made by using JAVA language that has advantages in developing graphical user interface. This software can perform feature extraction by using multi derivatives and Spearman's correlation with an arbitrary dependent variable after the preprocessing of the spectrum between 1000–2500 nm Wavelength by Standard Normal Variate. In addition, this software can visualize the tendency of the data by the PCA method, and determine the regression model by the PLS method. We demonstrated the usability of this software using Japanese green tea samples. A set of ranked green tea samples from a Japanese commercial tea contest was analyzed by Fourier transform near-infrared (FT-NIR) reflectance spectroscopy. This FT-NIR data was analyzed by our software, and quality prediction model was made. This prediction model had enough high accuracy.
