Rapid and nondestructive identification of Belgian and Netherlandish Trappist beers by front-face synchronous fluorescence spectroscopy coupled with multiple statistical analysis

Abstract

Front-face synchronous fluorescence spectroscopy (FFSFS) was applied for the rapid and noninvasive recognition of Belgian and Netherlandish Trappist beers against non-Trappist beers. The front-face synchronous fluorescence spectra at wavelength intervals (??) of 30 and 60 nm for 80 bottles of beer, including 41 Trappist and 39 non-Trap-pist beers, were acquired in a 5 × 10 mm fused-quartz cuvette settled in a traditional right-angle sample compartment. The discrimination model was constructed by either principal component analysis (PCA) combined with linear discriminant analysis (LDA) or partial least squares-discriminant analysis (PLS-DA). Both PCA–LDA and PLS-DA models were validated by full (leave-one-out) cross-validation and k-fold cross-validation (k = 5). The PCA–LDA model presents reliable discrimination performance, with the cross-validated sensitivity (true positive rate) and specificity (true negative rate) in the range of 82.9–85.4% and 71.8–76.9%, respectively. The misclassification mainly occurs to a small portion of ambiguous Trappist and non-Trappist samples such as Abbey beers, which are rather similar to Trappist beers.