This work aims at comparing different non-targeted spectroscopic techniques (i.e., UV-Vis, FT-IR, FT-NIR, and NIR spectroscopy) for the authentication of white wine vinegar. Five white wine vinegars were adulterated with two different spirit vinegars. Further twenty-five wine vinegars were analysed to enlarge the authentic product dataset. All samples (i.e., 160) were analysed in duplicate by UV-Vis, FT-NIR, and FT-IR spectroscopy; moreover, a handheld NIR device was tested on a sub-set of samples (i.e., 89). Principal Component Analysis revealed sample patterns related to vinegar acidity (6 or 7.1%) rather than adulteration levels. After variable selection (SELECT algorithm), Linear Discriminant Analysis (LDA) models were developed and tested by independent external sets. The LDA models gave very high average correct classification rates in calibration (95.5-100.0%), cross-validation (92.4-100.0%), and prediction (90.0-100.0%) for all the spectroscopic techniques. With the portable NIR instrument, 100% correct classifications in prediction were obtained, demonstrating its suitability in vinegar authentication.
Spectroscopic Non-targeted Techniques in Combination with Linear Discriminant Analysis for Wine Vinegar Authentication / S. Grassi, C. Alamprese. - In: FOOD AND BIOPROCESS TECHNOLOGY. - ISSN 1935-5149. - 17:2(2024 Feb), pp. 479-488. [10.1007/s11947-023-03143-9]
Spectroscopic Non-targeted Techniques in Combination with Linear Discriminant Analysis for Wine Vinegar Authentication
S. GrassiPrimo
;C. Alamprese
Ultimo
2024
Abstract
This work aims at comparing different non-targeted spectroscopic techniques (i.e., UV-Vis, FT-IR, FT-NIR, and NIR spectroscopy) for the authentication of white wine vinegar. Five white wine vinegars were adulterated with two different spirit vinegars. Further twenty-five wine vinegars were analysed to enlarge the authentic product dataset. All samples (i.e., 160) were analysed in duplicate by UV-Vis, FT-NIR, and FT-IR spectroscopy; moreover, a handheld NIR device was tested on a sub-set of samples (i.e., 89). Principal Component Analysis revealed sample patterns related to vinegar acidity (6 or 7.1%) rather than adulteration levels. After variable selection (SELECT algorithm), Linear Discriminant Analysis (LDA) models were developed and tested by independent external sets. The LDA models gave very high average correct classification rates in calibration (95.5-100.0%), cross-validation (92.4-100.0%), and prediction (90.0-100.0%) for all the spectroscopic techniques. With the portable NIR instrument, 100% correct classifications in prediction were obtained, demonstrating its suitability in vinegar authentication.File | Dimensione | Formato | |
---|---|---|---|
s11947-023-03143-9.pdf
accesso aperto
Tipologia:
Publisher's version/PDF
Dimensione
2.52 MB
Formato
Adobe PDF
|
2.52 MB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.