Prediction of electrophoretic mobility of substituted aromatic acids in different aqueous–alcoholic solvents by capillary zone electrophoresis based on support vector machine

Abstract

The electrophoretic mobilities of 26 substituted aromatic acids in two different aqueous–alcoholic (ethanol and methanol) solvents in capillary zone electrophoresis were predicted based on support vector machine (SVM) using five molecular descriptors derived from the structures of the substituted aromatic acids, the dielectric constant of mixed solvents and the energy of the highest occupied molecular orbital of the methanol and ethanol. The molecular descriptors selected by stepwise regression were used as inputs for radial basis function neural networks (RBFFNs) and SVM. The results obtained using SVMs were compared with those obtained using the regression method and RBFFNs. The prediction result of the SVM model is better than that obtained by regression method and RBFFNs. For the test set, a predictive correlation coefficient R = 0.9974 and mean square error of 0.2590 were obtained. The prediction results are in very good agreement with the experimental values.