INVESTIGADORES
WUNDERLIN Daniel Alberto
artículos
Título:
Composition of Honey from Córdoba (Argentina): Evaluation of North-South Provenance by Chemometrics.
Autor/es:
BARONI, M.V.; ARRÚA, R.C.; NORES, M.L.; FAYE, P. F.; DÍAZ, M. P.; CHIABRANDO, G. A.; WUNDERLIN, D. A.
Revista:
FOOD CHEMISTRY
Editorial:
Elsevier
Referencias:
Lugar: Amsterdan; Año: 2009 vol. 114 p. 727 - 733
ISSN:
0308-8146
Resumen:
We report the characterisation of honey samples produced in Córdoba (Argentina) and their classification
by geographical provenance (North/South) using chemometrics. Twenty-two variables were analysed
considering both chemical properties and mineral profile. Honey samples were found to meet the international
specifications for the evaluated parameters.
Classification of honey in according to its geographical provenance (North/South) was achieved by pattern
recognition techniques applied to 15 out of 22 variables. Glucose, pH, free acidity, free amino acids,
calcium and zinc were selected by stepwise discriminant analysis, explaining the classification of honey
according to their geographical origin. Application of k-nearest-neighbour classification procedure to
these six selected variables produced a successful assignation (99% correct) of honey to its provenance.
On the other hand only 83% right assignation was observed, when the 15 variables were used, confirming
that the use of all available features is unnecessary to get good geographical discrimination.
these six selected variables produced a successful assignation (99% correct) of honey to its provenance.
On the other hand only 83% right assignation was observed, when the 15 variables were used, confirming
that the use of all available features is unnecessary to get good geographical discrimination.
these six selected variables produced a successful assignation (99% correct) of honey to its provenance.
On the other hand only 83% right assignation was observed, when the 15 variables were used, confirming
that the use of all available features is unnecessary to get good geographical discrimination.
k-nearest-neighbour classification procedure to
these six selected variables produced a successful assignation (99% correct) of honey to its provenance.
On the other hand only 83% right assignation was observed, when the 15 variables were used, confirming
that the use of all available features is unnecessary to get good geographical discrimination.