Quantitative structure-retention relationship (QSRR) analysis has been
performed in order to correlate the retention of selected estradiol
derivatives with their calculated molecular lipophilicity. The lipophilicity
descriptors were derived computationally and most important were selected.
Linear regression (LR) Wad used for model establishing. Statistical quality
of the generated models was determined by standard statistical and
crossvalidation statistical parameters. Statistically significant and
physically meaningful models were obtained. The prediction results are very
well correlated with the experimentally observed data. Given predictive
ability of the established models indicates that they could be used for
predicting the chromatographic behavior of the similar molecules in
normal-phase high-performance thin-layer chromatography. [Projekat
Ministarstva nauke Republike Srbije, br. 172012, br. 172014 i br. 31055]