Multivariate analysis of UV spectra of complex herbal mixtures and essential oils

Abstract

44 UV spectra of essential oils and plant extracts were analyzed in multivariate way by robust principal component analysis. Although the samples were not clustered against their chemical composition, an interesting dependence was found. Most of the oils had the first PC1 positive, other samples - negative, which was connected with an intensive band just above 200 nm and above 300 nm. These results suggest the ability to distinguish between oils and other extracts by discriminant analysis models, and first PC1 acts as a good discriminant.