Quadratic Discriminant Analysis Methods Research Articles

IDENTIFICATION OF SALMONELLA‐INOCULATED BEEF USING A PORTABLE ELECTRONIC NOSE SYSTEM

ABSTRACT A commercially available Cyranose‐320TM (Cyrano Sciences, Pasadena, CA) conducting polymer–based electronic nose system was used to analyze the volatile organic compounds emanating from vacuum‐packaged beef strip loins (Longissimus Lamborum) that were repackaged simulating retail store conditions and stored at 4 and 10°C after inoculating them with Salmonella Typhimurium. Two statistical techniques, i.e., linear discriminant analysis (LDA) and quadratic discriminant analysis (QDA), were used to develop classification models from the collected sensor signals. The models were developed for the captured signals, which were scaled and not scaled. The performances of the developed models were validated by two different methods, leave‐one‐out cross validation and bootstrapping. The developed models classified the meat samples based on the Salmonella population into “No Salmonella” (microbial counts < 0.7 log10 cfu/g) and “Salmonella” (microbial counts ≥ 0.7 log10 cfu/g). Overall, scaling the signals improved the classification accuracies obtained. The quadratic discriminant‐based classification models validated by the bootstrapping technique performed better than the LDA‐based models. For the meat samples stored at 10°C, the highest classification accuracy obtained by the QDA method with bootstrapping validation was 80.5% when the signals were scaled. For the samples stored at 4°C, the QDA method provided the highest classification accuracy of 87.3% using bootstrapping validation. The results obtained prove that the electronic nose system could identify meat samples contaminated with S. Typhimurium at a population concentration level of 0.7–2.6 log10 cfu/g.

SPOILAGE IDENTIFICATION OF BEEF USING AN ELECTRONIC NOSE SYSTEM

A commercially available Cyranose-320. conducting polymer-based electronic nose system was used to analyzethe volatile organic compounds emanating from fresh beef strip loins (M. Longisimmus lumborum) stored at 4C and 10C.Two statistical techniques, i.e., linear discriminant analysis (LDA) and quadratic discriminant analysis (QDA), were usedto develop classification models from the collected sensor signals. The performances of the developed models were validatedby two different methods: leave-1-out cross-validation, and bootstrapping. The developed models classified meat samplesbased on the microbial population into unspoiled (microbial counts <6.0 log10 cfu/g) and spoiled (microbial counts >6.0 log10 cfu/g). Overall, quadratic discriminant-based classification models performed better than linear discriminantanalysis based models. For the meat samples stored at 10C, the highest classification accuracies obtained by the LDAmethod with leave-1-out and bootstrapping validations were 87.10% and 85.87%, respectively. On the other hand,classification by QDA and subsequent validation by leave-1-out and bootstrapping provided highest accuracies of 87.5%and 97.38%, respectively. For samples stored at 4C, the LDA method provided highest classification accuracies of 79.17%and 85.64% using leave-1-out and bootstrapping validation, respectively. When the QDA method was used, the highestclassification accuracies obtained for the samples stored at 4C were 87.50% and 98.48%, respectively, with leave-1-outand bootstrapping validations.