Classification of monovarietal Italian olive oils by unsupervised (PCA) and supervised (LDA) chemometrics

This research work was conducted to establish whether monovarietal olive oils could be differentiated by their basic classes of compounds, ie (a) fatty acids, (b) fatty alcohols, (c) polycyclic triterpenes and (d) squalene. The ratio values of biosynthetically correlated acids were also examined. The mentioned classes of compounds, formed in distinct biosynthetic compartments of the olive fruit, should represent characteristic compositional data of an olive cultivar. The widely cultivated Italian olive cultivars studied were Frantoio, Bosana, Dritta and Leccino. Principal component analysis (PCA) was applied to the analytical data to reveal the compounds (variables) with the highest weights (loadings), with the aim of using them in subsequent computations. These variables were tetracosanol, hexacosanol Delta(5)-avenasterol, cycloartenol, 24-methylencycloartenol, oleic, linoleic, linolenic, stearic and palmitoleic acids and the ratios palmitic/stearic, palmitic/palmitoleic and linoleic/linolenic. Linear discriminant analysis (LDA), carried out on a training set of 57 oils (13 Dritta, 25 Leccino, 12 Frantoio and seven Bosana) produced a 96% correct group classification. The prediction LDA model created with the training set was validated with a test set of 19 oil samples (six Dritta, seven Leceino, four Frantoio and two Bosana), permitting accurate classification of all the 'unknown' olive oils. (C) 2003 Society of Chemical Industry.