The analysis of sensory and chemical–physical variables via multivariate additive PLS splines

When the information given by a panel of judges on wine sensory characteristics is resumed in a three-way table, Y, and the objective is to extract knowledge from a chemical data-set, X, which has a predictive power for wine sensorial variables, the study of the relationships between chemical–physical variables and sensorial data sets is really complex. In this paper after resuming the information in Y, we propose to consider multivariate additive partial least squares via splines, recently presented in literature, to predict Y by X.

► The study of the relationships among sensory and chemical–physical variables.
► Multivariate additive partial least squares via splines.
► Main chemical–physical predictors and Interaction chemical–physical predictors.
► Generalised cross validation criterion.