Multivariate modeling for detecting adulteration of extra virgin olive oil with soybean oil using fluorescence and UV–Vis spectroscopies: A preliminary approach

Abstract

This work presents a comparative study of chemometric methods used to quantify adulteration of extra virgin olive oil (EVOO) with soybean edible oil using fluorescence and UV–Vis spectroscopies. The adulteration was prepared by adding soybean edible oil in different concentrations (10, 50, 100, 150, 200, 250 and 300 g/kg). Different multivariate regression strategies were evaluated: partial least squares (PLS) using full spectrum; PLS with significant regression coefficients selected by the Jack-Knife algorithm (PLS-JK) and multiple linear regression (MLR) with previous selection of variables by stepwise algorithms (SW-MLR); successive projections algorithm (SPA-MLR); and genetic algorithm (GA-MLR). The predictive ability of the models was assessed, for each spectroscopic technique. For fluorescence spectroscopy, satisfactory prediction results were obtained for all the regression models with Root Mean Square Error of Prediction (RMSEP) values varying from 14.0 to 17.5 g/kg. When the regression methods were evaluated for UV–Vis spectra, higher RMSEP values were found, varying from 13.3 to 30.4 g/kg. The results indicate that the two spectroscopic techniques have similar performances with respect to predictive ability of the regression models.