2017 Volume 18 Issue 2 Pages 87-92
This review provides an overview of the data mining on fluorescence fingerprint (FF) and the application of data mining to flow cytometry (FCM). Canonical discriminant analysis, which is one of data mining method, on FF of taros revealed that their geographic origins (Japan or China) could be predicted as accurate as the standard methods such as inorganic elements composition and isotopic ratio analyses. In addition, the histogram of the fluorescence signal obtained by FCM for the prediction of Escherichia coli (E. coli) concentrations in green tea beverage were analyzed with partial least squares regression (PLSR) method, and it was clarified that E. coli concentration could be predicted by PLSR with higher accuracy than conventional FCM with reduced false positives and false negatives. Data mining can be applied to both FF and metabolomics for sample quality estimation or exploration of important compounds. Therefore, using data mining as a clue, it would be expected that the development of research that integrates or compares FF and metabolomics data for the understanding of overall food quality in more detail.