Analysis of different Flos Chrysanthemum tea samples with the use of two-dimensional chromatographic fingerprints, which were interpreted by different multivariate methods

Abstract

Flos Chrysanthemum tea contains flavonoids, essential oils and caffeoylquinic acids. These substances are pharmacologically active but this activity is cultivar dependent. Seventy-six Flos Chrysanthemum samples collected from four cultivars (Hangju, Taiju, Gongju and Boju) were discriminated with the use of results from high performance liquid chromatography (HPLC) and gas chromatography-mass spectrometry (GC-MS). A two-dimensional chromatographic fingerprint dataset of the four kinds of Flos Chrysanthemum cultivar was built from the combined GC/HPLC profiles and thirty variables were selected. Principal component analysis (PCA) and kernel – PCA (KPCA) were used for feature extraction. The score mapping graph indicated that these two PCA methods effectively extracted most information from the samples, and the four Flos Chrysanthemum cultivars were qualitatively differentiated. Furthermore, four supervised pattern recognition techniques, radial basis function-neural network analysis (RBF-NN), least squares support vector machines (LS-SVM), linear discriminant analysis (LDA) and K-nearest neighbors (KNN), successfully predicted the validation of the samples.