Abstract 014
Application of Pattern Recognition Techniques to Essential Oil of Hops
IUFOST Symposium on Food Research and Data Analysis, Oslo, 1982 pp. 424.
K.J. Siebert, L.E. Stenroos, and M.C. Meilgaard
Some 54 hop samples representing nine different varieties grown in Europe and the U.S. were obtained. The hop samples were steam distilled in order to collect the oils, which were then subjected to capillary column gas chromatography. Fifty-two peaks were measured in each chromatogram. The results were examined by Stepwise Discriminant Analysis (BMDP7M) and by several of the ARTHUR routines. Selection of a limited number of parameters was done to minimize the problem of an almost equal number of samples and measurements. The U.S.-grown Tettnang sample was grouped with the European grown hops and with the North American-grown hops in separate Discriminant Analyses. In both cases, the program succeeded in developing non-overlapping classification functions. Examination of the differences in the parameters chosen gave insight into the oil components which indicate place-of-growth and those which are more dependent on variety. The data were also grouped according to the five varieties for which there were a substantial number of samples. Discriminant Analysis had difficulty in resolving all the groups with the concentration basis data, but succeeded with the relative (% of the total oil) basis results, especially the myrcene corrected set. A number of ARTHUR procedures were also successful in separating the varietal groups, particularly MULTI, LEAST, STEP and the SIMCA routines. Here, too, the relative basis data set appeared to be slightly easier to resolve. It appears possible to discern whether a hop sample was grown in Europe or North America through the examination of its essential oil pattern. In a similar manner, the variety of a hop sample can be estimated, although with less certainty.
Back to K.J. Siebert home page | Back to publication list