prediction of wine quality and geographic origin from chemical measurements by parital least-squares regression modeling

A multivariate regression method, PLS, was applied to model the relationship between objective chemical measurements and subjective sensory evaluation of Pinto Noir wine samples. Descriptive and predictive models were calculated according to present pathways in order to classify the wines according to their geographic origin and to predict several organoleptic characteristics. The importance of inorganic elements in these prediction problems was investigated.