Classification of Polish wines by application of ultra-fast gas chromatography
- 31 Downloads
The potential of ultra-fast gas chromatography (GC) combined with chemometric analysis for classification of wine originating from Poland according to the variety of grape used for production was investigated. A total of 44 Polish wine samples differing in the type of grape (and grape growth region) used for the production as well as parameters of the fermentation process, alcohol content, sweetness, and others which characterize wine samples were analysed. The selected features coming from ultra-fast GC analysis were subsequently used as inputs for both principal component analysis (PCA) and supervised machine learning. Using the proposed classification algorithm, it was possible to classify white and red wines according to the variety of grape used for production with a 98.7 and 98.2% accuracy, respectively. The model was characterized by good recall and area under receiver operating characteristic which was 1.000 for white wines and 0.992 for red wines. Cuveé wines (made from various types of grapes) were also successfully classified which leads to the conclusion that the proposed classification method can be used not only to differentiate between wines made from different grapes but also to detect possible adulterations, provided known; non-adulterated samples are available as a reference. The model was also used to classify wine samples based on other features, such as the geographic region in which the vineyard is situated, type of yeast used, the temperature of fermentation, sweetness, etc. In all cases, a high classification accuracy (in most cases > 90%) was achieved. The obtained results could be applied in the wine industry.
KeywordsWine Classification Ultra-fast gas chromatography Chemometric analysis Principal component analysis Support vector machines
Compliance with ethical standards
Conflict of interest
Justyna Płotka-Wasylka has received research mini-grant from by the Faculty of Chemistry, Gdańsk University of Technology. All authors declare that they have no conflict of interest.
Compliance with ethics requirements
This article does not contain any studies with human participants or animals performed by any of the authors.
- 3.Boser BE,. Guyon IM, Vapnik VN (1992) A training algorithm for optimal margin classifiers. In: Proceedings of the fifth annual workshop on Computational learning theory - COLT’92, pp. 144–152Google Scholar
- 5.Demšar J, Curk T, Erjavec A, Hočevar T, Milutinovič M, Možina M, Polajnar M, Toplak M, Starič A, Stajdohar M, Umek L, Zagar L, Zbontar J, Zitnik M, Zupan B (2013) Orange: data mining toolbox in Python. J Mach Learn Res 14:2349–2353Google Scholar
- 18.Sotirchos DG, Danezis GP, Georgiou CA (2017) Introduction, definitions and legislation. In: Georgiou CA, Danezis GP (eds) Food authentication: management, analysis and regulation. Wiley-Blackwell, Greece, pp 3–18Google Scholar