The oil extracted from the seeds of niger (Guizotia Abyssinica), collected from 6 different regions of Ethiopia and India, was characterized in terms of its fatty acid, sterol and triglyceride distribution and of its total tocopherol content. Where available, the results have been compared with those reported in the literature or with data on oils from the same botanical family (Compositae). The analytical data have then been elaborated by supervised pattern recognition techniques (Linear Discriminant Analysis (LDA) and Artifical Neural Network (ANN)) in order to authenticate the geographical origin of the samples. Eight and 11 variables were necessary to achieve a complete discrimination respectively of the country and of the region of origin of the oils under exam, when using LDA, whereas ANN required a smaller number of experimental variables (4 and 6), due to its non-linearity.
Characterization of the lipid fraction of Niger seeds (Guizotia abyssinica cass.) from different regions of Ethiopia and India and chemometric authentication of their geographical origin / Marini, Federico; Magri', Antonio; Domenico, Marini; Fabrizio, Balestrieri. - In: EUROPEAN JOURNAL OF LIPID SCIENCE AND TECHNOLOGY. - ISSN 1438-7697. - STAMPA. - 105:11(2003), pp. 697-704. [10.1002/ejlt.200300797]
Characterization of the lipid fraction of Niger seeds (Guizotia abyssinica cass.) from different regions of Ethiopia and India and chemometric authentication of their geographical origin
MARINI, Federico;MAGRI', Antonio;
2003
Abstract
The oil extracted from the seeds of niger (Guizotia Abyssinica), collected from 6 different regions of Ethiopia and India, was characterized in terms of its fatty acid, sterol and triglyceride distribution and of its total tocopherol content. Where available, the results have been compared with those reported in the literature or with data on oils from the same botanical family (Compositae). The analytical data have then been elaborated by supervised pattern recognition techniques (Linear Discriminant Analysis (LDA) and Artifical Neural Network (ANN)) in order to authenticate the geographical origin of the samples. Eight and 11 variables were necessary to achieve a complete discrimination respectively of the country and of the region of origin of the oils under exam, when using LDA, whereas ANN required a smaller number of experimental variables (4 and 6), due to its non-linearity.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.