Application of pattern recognition to the discrimination of roasted coffees

Pattern recognition procedures have been applied to samples of roasted coffee. Some of them are torrefacto samples, that is coffee roasted with addition of sugar. Principal component analysis, cluster analysis, linear discriminant analysis, soft independent modelling of class analogies, and Spearman correlation studies have been carried out. Caffeine, aqueous extract, amino acids, polyphenols, 5-(hydroxymethyl)furfural, potassium, sodium, calcium, iron, manganese and magnesium have been used as chemical descriptors.