Usage of K-cluster and factor analysis for grouping and evaluation the quality of olive oil in accordance with physico-chemical parameters

Abstract

25 olive oils were studied– different in origin and ways of extraction, in accordance with 17 physico-chemical parameters as follows: color parameters – a and b, light, fluorescence peaks, pigments – chlorophyll and β-carotene, fatty-acid content. The goals of the current study were: Conducting correlation analysis to find the inner relation between the studied indices; By applying factor analysis with the help of the method of Principal Components (PCA), to reduce the great number of variables into a few factors, which are of main importance for distinguishing the different types of olive oil;Using K-means cluster to compare and group the tested types olive oils based on their similarity. The inner relation between the studied indices was found by applying correlation analysis. A factor analysis using PCA was applied on the basis of the found correlation matrix. Thus the number of the studied indices was reduced to 4 factors, which explained 79.3% from the entire variation. The first one unified the color parameters, β-carotene and the related with oxidative products fluorescence peak – about 520 nm. The second one was determined mainly by the chlorophyll content and related to it fluorescence peak – about 670 nm. The third and the fourth factors were determined by the fatty-acid content of the samples. The third one unified the fatty-acids, which give us the opportunity to distinguish olive oil from the other plant oils – oleic, linoleic and stearin acids. The fourth factor included fatty-acids with relatively much lower content in the studied samples. It is enquired the number of clusters to be determined preliminary in order to apply the K-Cluster analysis. The variant K = 3 was worked out because the types of the olive oil were three. The first cluster unified all salad and pomace olive oils, the second unified the samples of extra virgin oilstaken as controls from producers, which were bought from the trade network. The third cluster unified samples from pomace and extra virgin oils, which distinguish one from another in accordance with their parameters from the natural olive oils, because of presence of plant oils impurities.