Vedecký časopis - archív

49 2010

Journal of Food and Nutrition Research
Súhrny čísla 2 / 2010

KOREŇOVSKÁ, M. – SUHAJ, M.
Assessment of some European goats’ cheeses by multivariate elemental data comparison
Journal of Food and Nutrition Research, 49, 2010, č. 2, s. 94-103

Milan Suhaj, VÚP Food Research Institute, Department of Chemistry and Food Analysis, Priemyselná 4, SK – 824 75 Bratislava 26, Slovakia. E-mail: suhaj@vup.sk, tel.: +421-2-50237146, fax: +421-2-55571417

Súhrn: Based on statistical evaluation of the contents of 12 mineral and trace elements (Ba, Ca, Cu, Cr, Hg, K, Mg, Mn, Mo, Na, Ni and V) determined by atomic absorption spectrometry, differentiation and classification of 158 European commercial caprine hard cheeses originating from 10 countries were carried out involving the cluster, principal component, factor and discriminant analyses. Cluster analysis grouped the samples into several sub-clusters in principle corresponding well with the country or region of sample origin. Similar clustering tendency of samples was revealed also by principal component and factor analyses. The best classification of caprine cheeses according to their country of origin was achieved by K-th nearest neighbour discrimination, with 100% correctly classified samples at K = 1, and 87.9% correctness in cross-validation tests. When processed by canonical discriminant analysis, 87.5% recognizability and 80.4% successfulness in cross-validation tests was achieved. Results obtained indicated that caprine cheeses originating from Germany, Italy, Croatia, Spain and Portugal could be 100% correctly traced according to their country of origin on the basis of multivariate analysis of elemental markers. In this context, Cr, Mo, Mn, Hg, Ca, Na, V and Ba were recognized as the most significant elemental markers providing the highest discrimination power for caprine cheeses geographical distinguishing.

Kľúčové slová: caprine cheeses; elemental data; multivariate analysis; classification

Na stiahnutie:
  jfnr-2010-2-p094-103.pdf (PDF, 197.3 Kb, 1656x)