APPLICATION OF MULTIVARIATE MATHEMATICAL STATISTICAL-METHODS FOR THE COMPARISON OF THE RETENTION BEHAVIOR OF POROUS GRAPHITIZED CARBON AND OCTADECYLSILICA COLUMNS

The retention of sixteen ring-substituted aniline derivatives on porou s graphitized carbon (PGC) column (eluents: acetonitrile-water and met hanol-water mixtures) and on octadecylsilica (ODS) column (eluents: me thanol-0.025 M K2HPO4 mixtures) was determined, and the relationship b etween retention characteristics and physico-chemical parameters were evaluated by principal component analysis followed by two-dimensional non-linear mapping and by cluster analysis and also by canonical corre lation analysis. Calculations proved that marked differences can be de tected between the retention characteristics of PGC and ODS columns, a nd the electronic and steric parameters of aniline derivatives have a great impact on their retention. The comparison of various multivariat e mathematical-statistical methods indicated that principal component analysis followed by two-dimensional non-linear mapping is the most ap propriate method for evaluation of large data matrices in reversed-pha se liquid chromatography.