Application of hierarchical classification models and reliability estimation by bootstrapping, for authentication and discrimination of wine vinegars by UV–vis spectroscopy

VinegarScan: A Computer Tool Based on Ultraviolet Spectroscopy for a Rapid Authentication of Wine Vinegars

“…Nevertheless, among the aforementioned strategies that have enabled the characterization and classification of them according to their origin or category, UV-vis spectroscopy has demonstrated in the last few years its great ability to characterize and classify wine vinegar according to their quality, particularly those with a protected designation of origin (PDO), and to their production process or aging time, showing successful results and opening up the possibility of developing an efficient quality control tool [15][16][17][18][19]. UV-vis spectrum of wine vinegars, and also wines, presents absorption bands in the range 300-400 nm related to absorbent species relevant for color and taste characteristics, such as polyphenolic compounds (e.g., some acids as hydroxycinnamic acids, some aldehydes such as benzaldehyde, stilbenes, flavanols, and anthocyanins), together with the presence of other compounds such those related to the grape-must caramel added to the vinegars [17,18]. This makes a unique fingerprint that can be characteristic for each type or category of vinegar.…”

“…In particular, a previous study [17] showed the classification ability of UV-vis spectroscopy to differentiate many different wine vinegars according to different categories and to characterize wine vinegars without PDO with different ages and origins. Moreover, there is also a recently published study that reaffirms the ability of differentiation PDO wine vinegars by UV-vis spectroscopy [19].…”

“…Among the main chemometric techniques for classification purposes, partial least square-discriminant analysis (PLS-DA), decision trees (DT), and artificial neural networks (ANN) are the most highlighted [14,[20][21][22][23]. In a previous UV-vis study of vinegars [17], a hierarchical classification model (HCM) formed by soft independent modelling-class (SIMCA) and partial least squares-discriminant analysis (PLS-DA) models in each of the nodes was needed to classify all the vinegar samples due to many different characteristics or vinegar types were considered (i.e., aged or not aged, with or without a PDO, from different PDOs and for different aging categories). HCM was needed for this study due to the classification problem being complex, since one or more classes can be divided into subclasses or grouped into superclasses, that is, with the classes to be predicted being hierarchically related.…”

“…Despite the successful results obtained [17], transferring the developed methodologies to the productive and industrial sectors for routine analysis remains challenging. Nonexperts in data treatment and winemakers or workers of food companies or industries, which are very interested in the spectroscopic quality control of their products, are not always familiar with pre-processing, data modelling, and in general, chemometrics.…”

“…Due to the promising results acquired in the previous study [17], together with the simplicity, price, speed, and portability of the UV-vis spectrometer, the objective of this work was the development of a new simple and easy-to-use software that, combined with a portable UV-vis spectroscopic device, could be able to authenticate and classify wine vinegars in a considerably short time and friendly-use way. It could be implemented as a routine method for process control and monitoring (directly in-situ and at real-time) in the wine vinegar production chain.…”