Identifier: TFG:1203
Authors: Obisesan, Kudirat Abidemi
Abstract:
Identification of the geographical origin of palm oil by multivariate techniques The interest of the consumers is growing over the origin indication of food products, as well as health benefits, which are related to the geographical origin. The main objective of this study is the differentiation and classification of palm oil samples from three different sources. The samples were analyzed at the University of Granada by HPLC, High Performance Liquid Chromatography Fingerprint technique with charged aerosol detector (CAD) and ultraviolet-visible (UV). Previous studies have evaluated the content of phytosterols as a possible oil variable differentiating, and it will be used in this work. Classification techniques were evaluated by SIMCA (Soft Independent Modelling of Class Analogy) and PLS-DA (Partial Least Squares Discriminant Analysis). For the establishment of the classification, various studies, including the discrepant samples, pretreatment, the discriminating region and latent variables were performed to obtain the optimum working condition. The presence of the discrepant samples had an influence on the results of CAD. In CAD and UV data, two groups of samples are observed in the Principal component analysis, which are Asia and America. However samples of Africa are dispersed by the two continents. Like any analysis technique, once the model its established, it can be validated. The validation of multivariate classification models is widely accepted as quality parameters, the values of percentages of: true positives (TP), true negative (TN), false positive (FP) and false negatives (FN). These parameters give an idea of how well the model. PLS-DA-CAD, it was the one with the best results on the designation of origin of the palm oil samples with positive real rates of 96%, 86%, 73% for class 1, 2 and 3 respectively.
Repositori URV