Selection of near-infrared wavelengths for monitoring milk coagulation using principal component analysis

Principal component analysis was used for the selection of near infrared wavelengths for monitoring the change in macropeptide concentration of milk during the enzymatic phase of coagulation. The selection was based on the plot of the second principal component loading, the potential for monitoring phenylalanine and methionine, and the physical magnitude of the change in reflectance during coagulation. The wavelengths chosen were 1250, 1450, 1650, 1750, 1800, and 1940 nm. A multiple linear regression model was developed which correlated changes in macropeptide concentration with the first and second principal component (PC1 and PC2) scores using six selected wavelengths. This model was found to be as effective in describing the variation of the macropeptide concentration as a model that used the entire spectrum. A regression model developed using six spectral bands (±8 nm) at the previously specified peak wavelength showed that the PC1 and PC2 components of the six spectral bands were as effective for monitoring the macropeptide concentration of coagulating milk as the entire spectrum.