Prediction of solubility parameters for polymers by a QSPR model

A set of molecular descriptors, such as hb, alk, E-int, Q(ii), Q(H) and n(N), which are obtained directly from polymeric repeating unit structures, are used to predict the solubility parameter (delta) of polymers and generate a quantitative structure-property relationship (QSPR) model, with the standard error s of 0.75 (J/cc)(0.5) for the training set of 51 polymers and the standard error s of 1.01 (J/cc)(0.5) for the test set of 46 polymers. The calculated results show that the QSPR model is easier to apply and has better predictive capability than the existing models.