A PLSR Model for Consumer Preference Prediction of Yoghurt from Sensory Attributes Profiles

Abstract Consumer preference investigations are extremely emphasized in marketing management strategy for food enterprises. In this communication, a partial least squares regression (PLSR) model for consumer preference prediction from sensory attributes profiles is developed and three brands of yoghurt are exampled to demonstrate the accuracy and practicability of the method. The model also provides the importance ranking of the analysed sensory variables, which may give a guidance for industrial production and sensory experiment design. Drawing on the results of PCA and hierarchical clustering analysis, the variable filtering process of PLSR is proved to be theoretically correct.