A principal component analysis approach for developing retention models in liquid chromatography

Three retention models for liquid chromatography are developed using principal component analysis (PCA). It is shown that they exhibit features similar to that of the model based on linear solvation energy relationship (LSER). However, the fitting performance of the PCA models is better than that of the LSER model, the performance of which can be considerably improved by the use of artificial neural networks. In addition, the possibility of using the proposed models as well as the LSER model to predict the retention times of solutes under chromatographic conditions at which these solutes have never been studied is also examined by means of three data sets of analytes consisting of non-polar compounds to polar compounds with a variety of functional groups.

► Three new retention models for liquid chromatography are developed using PCA.
► The proposed models are successfully validated using a great variety of analytes.
► They can predict the retention times of solutes without their previous study.
► The accuracy of the obtained results is good and better than that of LSER models.