Prediction of key sensory attributes of apple juices by multivariate analysis of their physicochemical profiles

Abstract

Purpose The purpose of this paper is to analyse physicochemical and sensory profiles of apple juices using multivariate methods and to predict the essential sensory properties by means of instrumental measurements. Design/methodology/approach The studied set included apple juices reconstituted from concentrate (clear and cloudy) and direct (pasteurised and freshly squeezed, not pasteurised). The physicochemical (colour, turbidity, total soluble solids, sucrose, D-glucose, D-fructose content, pH, total acidity, total phenolic, total flavonoid content and antioxidant capacity) and sensory profiles of juices were analysed by principal component analysis and partial least squares (PLS) regression. Findings An exploratory study of physicochemical and sensory profiles revealed dimensions differentiating apple juices. The first dimension was related to their physical properties (clarity, sediment and colour coordinates L*, C*, h°), whereas the second dimension referred to their flavour, being positively correlated with sourness and astringency, and negatively with sweetness. The PLS regression analysis revealed the quantitative relationship between sweet and sour flavours and the instrumental measurements. The physicochemical parameters of apple juices contributing positively or negatively to the key sensory attributes were determined based on the regression coefficients. Originality/value This study presents a possibility to predict the key sensory attributes of apple juice on the basis of selected physicochemical parameters. This approach could be a supporting tool for sensory analysis that could allow the quality characterisation on a large sample set and hence shorten the time necessary for routine quality control, product design and development.