Study of some European cheeses geographical traceability by pattern recognition analysis of multielemental data

Pattern recognition techniques (cluster, principal component, factor and discriminant analysis) were performed for tracing the geographical origin of some Emmental and Edam type hard cheeses originated from ten European countries. Two hundred and seventy-eight cheese samples were analysed on the content of elemental markers Ba, Ca, Cr, Cu, Hg, K, Mg, Mn, Mo, Na, Ni and V by atomic absorption spectrometry. Cluster analyses showed some preliminary results intimated partial grouping of cheeses according to country of origin. Principal component and factor analysis showed to be effective methods for data segregation and visualization. Canonical discriminant analysis demonstrated 85.9 and 93% potential to distinguish geographical differences and correct classification of cheeses according to multielemental data collected during the 2 years of production. Cross-validation of two testing data sets led to 86.4 and 90% correctness of cheeses classification. The stepwise discriminant analysis arranged the markers according to their influence on the discrimination procedure in descending order: Na, Cr, Cu, Ba, Ca, Hg, V, and Mg.