With regard to the dummy regression, 5 PLS components were
retained in the model since after that the standard error of cross
validation no longer meaningfully decreased. The scores corresponding
to 5 PLS components could successfully discriminate 100% of the
subcutaneous fat samples according to the diet that the beef cows
were fed (hay or barley silage alone or combined with flaxseed)
(Fig. 4). Statistically significant differences (pb0.001) in some of the
studied FA between the subcutaneous fat samples from beef cows fed
diets with and without flaxseed could have
provided the basis for successfully classifying the whole sample set
according to the spectral data.