Modelling the relation between CieLab parameters and sensory scores for quality control of red-wine colour

Abstract

A linear relationship between certain physicochemical measurements and sensory evaluation of the colour of young red Rioja wines has been constructed. Among these physicochemical measurements, different variable selection procedures based on genetic algorithms and decision trees, showed that it suffices to take into account the CieLab parameters. With these variables and a data set of wines from the 1992 harvest, a linear formula was constructed by the partial least squares regression method, capable of correctly assigning 162 out of 170 samples from the 1993 harvest (error rate 4.6%). Thus, it can be used as a routine control for the annual problem of qualification of the new wine as young red Rioja wine.