基于主成分和聚类分析区分牡丹皮药材的差异性

目的：采用主成分和聚类分析对16批不同产地的牡丹皮药材进行分析，比较不同批次之间的差异。方法：选取每批牡丹皮药材色谱图中峰面积＞总面积1%的色谱峰作为变量，用SPSS 13.0软件进行主成分和聚类分析。结果：根据色谱图提取出3个主成分，累积贡献率84.011%，可基本描述30个变量的变异情况，不同产地的牡丹皮实现了良好的区分，共聚成四类。主成分分析和聚类分析的结果与牡丹皮药材产地有较好的一致性。结论：与指纹图谱相比，主成分和聚类分析对药材差异的区分更有效。

Discrimination of Moutan Cortex by principal component analysis and cluster analysis

Objective.. To analyze and discriminate 16 batches of Moutan Cortex from different regions by principal component analysis and cluster analysis. Methods： In chromatograms of Moutan Cortex,the peaks whose area was larger than 1% of the entire area were chosen as variables. SPSS 13.0 software was used for principal component analysis and cluster analysis. Results： Three principal components whose accumulative contribution rate reached 84. 011% were selected. Variation status of the 30 variables could be explained by the three principal component factors. Moutan Cortex from different regions were classified into four categories. The results of the principal component analysis and cluster analysis were consistent with the regions of Moutan Cortex. Conclusion： Compared with fingerprint, principal component analysis and cluster analysis were more useful for discrimination of crude materia medica from different regions.