Study of the phenolic profile of cider apple cultivars at maturity by multivariate techniques

Phenolic compounds in 46 Spanish cider apple varieties were determined by R P-HPLC with direct injection. Several pattern recognition procedures, inclu ding principal component analysis (PCA), linear discriminant analysis (LDA) , and partial least squares (PLS-1), were applied to the data in an attempt to classify the samples into bitter and nonbitter categories. Reliable dec ision rules were obtained by both LDA and PLS-1. LDA achieved 91.3 and 85.7 % correct classification respectively, for internal and external evaluation of the model.