Geographical origin identification of Chinese red wines using ultraviolet-visible spectroscopy coupled with machine learning techniques

Identifying geographical origins of red wines produced in specific regions is of great importance, since the geographical origins of wine influence its quality and price greatly. In this study, the feasibility of UV-Vis spectroscopy was evaluated for the classification of Chinese red wine samples according to their geographical origins, using principal component analysis (PCA) and two machine learning techniques: orthogonal partial least squares-discriminant analysis (OPLS-DA) and support vector machine (SVM). The PCA analysis indicated that there are differences in the chemical composition between wine samples from three different origins and inferred the chemical compounds responsible for the discrimination between wine geographical origins. Furthermore, OPLS-DA and SVM models were established to predict the class membership of wine samples from three different origins and results showed that both models can provide correct recognition rates of 100 % for wine samples in training and prediction sets. This study demonstrated that the combination of UV-Vis spectroscopy with machine learning-based modeling has the potential to be a simple, fast and low-cost tool for the routine identification of geographical origins of Chinese red wines.