Discrimination of wines Regularization Classification
Issue Date:
2007
Publisher:
Institute of Mathematics and Informatics Bulgarian Academy of Sciences
Citation:
Pliska Studia Mathematica Bulgarica, Vol. 18, No 1, (2007), 327p-339p
Abstract:
The method of regularized discriminant analysis (RDA) was used for identifying the geographical origin of wines on the base of chemical-analytical parameters in the scope of a European project "WINE DB"1. A data base with 63 measured parameters of 250 authentic wine samples from five countries of the vintage 2003 was taken as a basis for classifying and discriminating wines. Uni- and multivariate methods of data analysis were applied. By using a Matlab-program, which allows an interactive stepwise discriminant model building, some different models for authentic wines with corresponding classification and prediction error rates (resubstitution, classical and modified "Leave-one-out", simulation and test) will be presented. The goodness of our preferred model was analysed by classifying a test sample that was created by splitting the data set based on Duplex-algorithm of Snee. Project Steering Committee: R. Wittkowksi BfR, Germany, P. Brereton CSL, United Kingdom, E. Jamin Eurofins, France, X. Capron VUB, Belgium, C. Guillou JRC, Italy, M. Forina UGOA, Italy, U. Rmisch TUB, Germany, V. Cotea UIASI.VPWT.LO, Romania, E. Kocsi NIWQ, Hungary, R. Schoula CTL, Czech Republic.