ANALYSIS OF SENSORY AND INSTRUMENTAL DATA ON APRICOT PUREES WITH PATTERN-RECOGNITION TECHNIQUES

The correlation between sensory data and analytical parameters was studied using industrially produced apricot purees. The samples were characterized through different analytical parameters. Sensory evaluations were performed by a trained panel with respect to sweet, bitter, acid and whole taste. The correlation between sensory and analytical data was studied using the partial least-squares regression method. Gas chromatographic data regarding the volatile components were preliminary analysed by principal component analysis in order to select relevant, non-redundant compounds to be submitted to the regression analysis. Three significant relationships were obtained for sweet taste and two for acid taste, resulting in reductions of 41.1 and 30% in the X-block variance, respectively.