Application of Least-Square Support Vector Machines in Qualitative Analysis of Visible and Near Infrared Spectra: Determination of Species and Producing Area of Panax

Visible and near infrared (Vis&NIR) spectra of a sample can be treated as a signature, allowing samples to be grouped on basis of their spectral similarities. Vis&NIR spectra combined with LS-SVM have been used to discriminate species and producing area of panax. American Panax quinquefoliumI, Chinese Panax quinquefolium and Panax ginseng were analyzed in this study. Principal component analysis (PCA) was applied before LS-SVM modeling and the results indicated that the mass points of the spectral data (376-1025 nm) can be effectively reduced. For each model, half samples were used for calibration and the remaining half were used for prediction. The results of the PCA-LS-SVM models for discriminating the both species and producing area of panax samples can reach 100% correct answer rate. The results of this study show that Vis-NIR spectroscopy technique combined with PCA-LS-SVM is a feasible way for qualitative analysis of discriminating herb producing areas and species.