摘要：A total of 40 commercial lager beers (6 dark, 28 pale and 6 alcohol free) were analysed by High-Performance Liquid Chromatography (HPLC) and evaluated by a descriptive sensorial panel. Discriminant Analysis was applied to 15 sensory descriptors to obtain a classification by types of beer. Cluster Analysis identified 4 clusters grouping the 15 sensory descriptors. The Correlation Matrix confirmed a correlation between them and the subsequent Principal Component Analysis. A stepwise discriminant analysis was used to eliminate the less significant descriptors, leading to 100% of the samples being correctly classified with a reduced number of variables (colour intensity, smell of caramel, smell of toasted, persistence and viscosity). The analysis of the correlations between iso-α-acid concentrations and sensory descriptors shows a good relationship with bitter taste. Thus, HPLC data can be used for predicting lager beer bitterness through the mathematical correlation developed in this paper.
