Support vector regression based QSPR for the prediction of some physicochemical properties of alkyl benzenes

Abstract

Physicochemical properties of alkyl benzenes are essential to separate pure component from alkyl benzene mixture. Support vector regression (SVR), a novel powerful machine learning technology based on statistical learning theory (SLT), integrated with topological indices was applied to the prediction of five physicochemical properties of alkyl benzenes including the normal boiling point (bp), enthalpy of vaporization at the boiling point ( H vb), critical temperature ( T c), critical pressure ( P c), and critical volume ( V c). In a benchmark test, SVR models for bp, H vb, T c, P c, and V c were compared with several modeling techniques currently used in this field. The prediction accuracy of the model was discussed on the basis of the leave-one-out cross-validation. The results show that the prediction accuracy of SVR model was higher than those of back propagation artificial neural network (BP ANN) and partial least squares (PLS) methods.