Stepwise Discriminant Analysis of Gas Chromatographic Data as an Aid in Classifying the Flavor Quality of Foods

TLDR: In this paper, a stepwise discriminant analysis for classifying food samples (known independently to differ in flavor) is illustrated by computer analysis of gas chromatograms from roasted coffee and potato chips.

Abstract: SUMMARY— Stepwise discriminant analysis for classifying food samples (known independently to differ in flavor) is illustrated by computer analysis of gas chromatograms from roasted coffee and potato chips. Four lots of coffee prepared so as to differ in flavor were scored organoleptically, steam distilled, and the distillate examined gas chromatographically. By calculating all possible ratios among peak heights and subjecting these ratios to discriminant analysis, the coffee could be classified into the four flavor categories from the gas chromatographic data. The discriminant analysis procedure was set to select the ratio most critical in differentiating among the chromatograms, then move on to the next most efficient ratio until the samples were classified. The same thing was done for potato chips except headspace volatiles were used for the gas chromatographic analysis. Not only does the procedure described enable flavor to be correlated with gas chromatographic data, but the efficiency values for each ratio are useful. A compound devoid of flavor can conceivably be highly correlated with flavor; however, there is a good chance that a compound highly correlated with flavor is a flavor substance itself.