Skip to main content



Abstract 113

Using chemometrics to classify samples and detect misrepresentation

in Progress in Authentication of Food and Wine, Ebeler, S.E., Takeoka, G.R. and Winterhalter, P., Eds. 2011, pp. 39-65

K.J. Siebert

 

Multivariate pattern recognition offers a number of advantages in detecting adulteration or misrepresentation of foods and beverages or their ingredients. Since multiple properties are used to make classifications, multiple adjustments would be necessary to perpetrate a successful fraud, which would likely make it uneconomical. Pattern recognition procedures can be either unsupervised (depending only on the structure of the entire data set) or supervised (using the presumed sample class identifications to establish classification rules). Unsupervised methods include principal components analysis (PCA) and cluster analysis. Supervised methods include nearest neighbor analysis, discriminant analysis, SIMCA and partial least squares discriminant analysis (PLS-DA) among others. The degree of success of classification rules produced should be tested by some validation procedure. Examples of pattern recognition applied to beer brands and hop cultivars are given in detail. Other applications are reviewed.

 

 

Back to K.J. Siebert home page | Back to publication list

Pages