The object of this study was to explore the feasibility of support vector machine (SVM) to identify the origin of chondroitin sulfate (CS) by near infrared spectroscopy. 96 batches CS from three different origins were collected in this research, 66 batches of which were chosen for training set by Kennard-Stone (KS) method and the rest were used for testing. Through the comparison of pretreatment methods of standard normal variate transformation (SNV), multiplicative scatter correction (MSC), Savitzky-Golay (SG) smoothing and derivative with SG smoothing, a SVM discrimination model was constructed, of which the prediction result is 100% accurate with the pretreatment of first derivative and SG smoothing with 15 points. The result indicated that it had great potential using SVM to identify the origin of CS.
Copyright © 2011 Elsevier B.V. All rights reserved.