setrchi.blogg.se

In addition, the following machine learning methods have been used to analyze volatile compounds in white wines: SVM, random forest (RF), multilayer perceptron (MLP), kNN and naive Bayes (NB). SVM and directed acyclic graph (DAG) decision tree have been applied to explore the volatiles’ fingerprints of red wines. Alternative data mining based on machine learning (ML) algorithms has a high potential for varietal authentication.

Non-targeted analysis of volatile compounds in combination with conventional statistical methods, such as principal component analysis (PCA), hierarchical component analysis (HCA) and linear discriminant analysis (LDA), has been employed for varietal differentiation of red wines. Forty important monoterpenes are found in grapes, including the following monoterpene alcohols and oxides: geraniol, linalool, citronellol, nerol, ( E)-hotrienol and cis- or ( Z)-rose oxide, which have floral aromas. These compounds include monoterpenes, C13-norisoprenoids, C6-compounds, methoxypyrazines and mercaptans. Some volatile compounds synthesized in grapes exist in volatile forms but mostly are non-volatile aroma precursors, which are released through biochemical and chemical reactions during fermentation and aging. The compounds derived from grapes provide varietal differentiation. The volatile compounds determining the aroma of wine originate from grapes (varietal aromas) and are secondary products of fermentation processes (fermentation aromas) and aging (post-fermentation aromas). It can be detected by volatile compound analysis. Varietal adulteration of wine is defined as the addition of must produced from grape varieties other than the labelled variety in illegal quantities. Moreover, the best SVM model (F-score of 1) was built with a subset containing 2-phenylethyl acetate and 3-(methylsulfanyl)propan-1-ol. An evaluation of the importance value of subsets consisting of six volatile compounds with the highest potential to distinguish between the Zweigelt and Rondo varieties revealed that SVM and kNN yielded the best classification models (F-score of 1, accuracy of 100%) when 3-ethyl-4-methylpentan-1-ol or 3,7-dimethyl-1,5,7-octatrien-3-ol (hotrienol) or subsets containing one or both of them were used.

Both machine learning methods yielded models with the highest possible classification accuracy (100%) when the relative concentrations of all the test compounds or alcohols alone were used as input data.

The relative concentrations of volatiles were used as an input data set, divided into two subsets (training and testing), to the support vector machine (SVM) and k-nearest neighbor (kNN) algorithms. 3,7-dimethyl-1,5,7-octatrien-3-ol (hotrienol) was found to be a variety marker for Zweigelt wines, since it was detected in all the Zweigelt wines, but was not present in the Rondo wines at all. Sixty-seven volatile compounds were tentatively identified in the test wines they represented several classes: 9 acids, 24 alcohols, 2 aldehydes, 19 esters, 2 furan compounds, 2 ketones, 1 sulfur compound and 8 terpenes. The wines were produced by using five commercial yeast strains and two types of malolactic fermentation. The aim of this study was to determine volatile compounds in red wines of Zweigelt and Rondo varieties using HS-SPME/GC-MS and to find a marker and/or a classification model for the assessment of varietal authenticity.