PRINCIPAL COMPONENT ANALYSIS (PCA) APPLIED TO NEAR-INFRARED SPECTRA FOR CLASSIFYING WHEAT FLOURS

The method was developed to perform principal component analysis (PCA) using widely used NECTM PC on near infrared spectra taken by IBMTM PC . As the application of this method, wheat flours with different proce ssing qualities were classified. Four kinds of commercially available wheat flours such as bread making four, Chinese noodle making flour, J apanese noodle making flour, and confectionery making flour were analy zed. The NIR spectra of 18 samples were recorded from 1100 to 2500 nm and the d(2) log (1/R) at several informative wavelengths were selecte d as variables for PCA. Principal components were calculated using the 11 d(2)log (1/R) values at the wavelengths where standard deviation o f d(2) log (1/R) were large and downward peaks of d(2) log (1/R) were observed. On the plane-with axes of the first and the third principal components, four kinds of wheat flours were clearly classified. On exa mination of the eigenvector of the first and the third principal compo nents, and the chemical constituents and physical properties of sample s, it can be thought that the first principal component relates to sam ple particle sizes, and the third one relates starch contents,. As a r esult, it can be concluded that near infrared spectroscopy has a possi bility to classify wheat hours with different processing qualities usi ng PCA.