Chemometrical investigation on four red wines from a single cultivar grown in the Piedmont region

The gas chromatographic profiles and phenolic contents of four red wines from a single cultivar grown in the Piedmont region were studied by multivariate feature selection and classification based on quadratic discriminant analysis. Of the 35 variables measured, four (condensed flavanoids, saturation, β-hydroxyethylbenzene and cis-hex-3-enol) were found to optimise the separation of the four wines: a classification rate of 100% and a prediction rate of 68.2% were obtained. Both volatile compounds and phenolic species were found to yield useful and complementary discriminant information.