Non-target Geographic Region Discrimination of Cabernet Sauvignon Wine by Direct Analysis in Real Time Mass Spectrometry with Chemometrics Methods

Direct analysis in real time integrated with high-resolution mass spectrometry (DART-HRMS) was developed in geographic region (i.e., altitude range and province) discriminating analysis of Cabernet Sauvignon wine by chemometrics methods. Based on eighteen features selected from merged mass spectra by correlation analysis and information gain, wines in different altitude ranges were discriminated by three chemometrics methods, including linear discriminant analysis (LDA), support vector machine (SVM) and Kohonen’s self-organising neural network (KohNN). The best result was obtained by KohNN classification models, which prediction accuracies were 96.26% and 93.33% for training set and test set, respectively. As for the provinces in China and France, KohNN classification models performed better, and its prediction accuracies were 95.04% and 91.49% for training set and test set, respectively. This study indicated that DART-HRMS combined with KohNN was a rapid efficient approach for wine geographic region discrimination.