Chemometrics
procedures [16,17] are usually applied to NIR spectra in
order to extract relevant information, as they are influenced by
the above mentioned features and other sources of error [18,19],
resulting that precision in NIR analysis is subjected to the
availability of algorithms or spectral data pre-treatments developed
to reduce these effects. It has been noted that the effect
of these unexpected variations in the spectra are particularly
important in the fat spectra versus those of other agro-food
products, since while the latter generally present very broad
absorption bands (more than 200 nm), fats and oils present very
sharp and narrow peaks [20].