Application of several statistical classification techniques to the differentiation of whisky brands

Abstract

Several supervised (stepwise discriminant analysis (SDA), statistical isolinear multi category analysis (SZMCA) and nearest neighbour analysis ( K N N ) ) and unsupervised (principal components analysis (PCA) and cluster analysis (CA)) classification techniques have been applied to analytical data jiom different whisky samples in order to distinguish between genuine whisky of a well known and expensive brand and other less expensive whiskies that could be used to replace the original contents of the bottle.

Varietal differences in the chemical composition and in vitro digestibility of rice straw

“…Overall hypothesis testing was by MANOVA (Vadiveloo 1989); stepwise discriminant analysis (Martin- Alvarez et al 1988) identified the more important variables from the range of variables analysed; these variables were subsequently ranked in the canonical variate analysis (Grant et al 1988). Using these variables as criterion variables, cluster analysis (Martin- Alvarez et al 1988;Blumenthal et al 1989;Herranz et al 1990;Vadiveloo & Fadel 1992) identified MR1 as a distinct variety and showed that within a variety, the main botanical fractions are similar in their chemical composition and IVD. Of the multivariate procedures used, cluster analysis is essentially an exploratory tool, since the cluster solution is not amenable to significance testing (Aldenderfer & Blashfield 1984).…”

Phenolic composition of black tea liquors as a means of predicting price and country of origin

“…Shimoda et al (1985) noted that the first and second principal components could discriminate between arabica and robusta coffee, differing blends, and degrees of roast. Martin-Alvarez et al(1988) could detect adulteration in whisky brands using PCA.…”