Quantifying binary and ternary mixtures of monovarietal extra virgin olive oils with UV–vis absorption and chemometrics.

The pigment profile of three monovarietal extra virgin olive oils (EVOOs) (Cornicabra, Picual, and Hojiblanca varietals) and their binary and ternary mixtures have been analyzed through visible spectroscopy. The information extracted from the registered spectra was treated and then modeled following two different chemometric approaches: a linear one based on multiple linear regression models, and a non-linear one based on the employment of artificial neural networks. All the designed models were validated using a k-fold cross-validation, and the largest mean absolute errors (MAEs) obtained for the varietal quantifications were 10% for the linear model and 2.8% for the non-linear one. These results let us prove the efficient generalization capability of these mathematical tools, as they are able to accurately quantify olive oil varietals in mixtures through their pigment profile only requiring visible spectroscopy data. [ABSTRACT FROM AUTHOR]