Inclusion of the standard deviation of data in principal component analysis. A graphical approximation

The adsorption capacity and specific adsorption surface area of 13 anti-hypoxia drugs were determined in three chromatographic systems using methanol-carbon tetrachloride, chloroform-carbon tetrachloride and acetonitrile-carbon tetrachloride mixtures as eluents. The retention behaviours of the anti-hypoxia drugs were compared using principal component analysis (PCA). A graphical approximation was used for the inclusion of the standard deviations of both the variables and observations in PCA and the results were visualized by two-dimensional nonlinear mapping and cluster analysis. The results indicated that the graphical approximation can be successfully used for the inclusion of the standard deviation of data in PCA calculations. Nonlinear mapping and cluster analysis resulted in similar, but not identical, classification of drugs and chromatographic systems, indicating that each multivariate method can be successfully used for the comparison of solutes and chromatographic systems.