VIEW ARTICLE    DOI: 10.1094/ASBCJ-47-0093

The Use of Multivariate Analysis of Beer Aroma Volatile Compound Patterns to Discern Brand-to-Brand and Plant-to-Plant Differences. K. J. Siebert and L. E. Stenroos, The Stroh Brewery Company, Detroit, MI 48207. J. Am. Soc. Brew. Chem. 47:0093, 1989.

A number of samples each of four beer brands produced at six different plants were taken at approximately weekly intervals. The beer aroma compounds were determined by Freon 112 extraction followed by capillary column gas chromatography. The results were examined using a number of multivariate analysis procedures. Some consistent patterns in the results were found that were partly or completely successful in characterizing either the beer brand or the plant of production. SIMCA, discriminant, and Bayesian analyses were particularly effective in deriving successful classification rules. Greater differences were observed between samples of different brands than between samples originating in different plants. It was of interest to note which compounds were important in making the distinctions observed and to consider whether they might result in significant flavor differences.

Keywords: Bayesian analysis, Cluster analysis, Discriminant analysis, Nearest neighbor analysis, Principal components analysis, SIMCA