Application of High-Order MCI Method to Predict Aqueous Solubility of Aliphatic Alcohols

Abstract

Correlations for estimation of the aqueous solubility (ln S) of aliphatic alcohols are proposed. The high-order MCI (molecular connectivity index) based quantitative structure-property relationship (QSPR) models are obtained by stepwise regression and support vector regression. On the basis of the data set of 50 aliphatic alcohols, the optimal linear model obtained by stepwise regression has a correlation coefficient of 0.990 and an average absolute error of 0.149 ln units and is comparable with the existing models. The optimal nonlinear model obtained by support vector regression has a correlation coefficient of 0.996 and an average absolute error of 0.116 ln units and is better than the existing models. The new models are predictive and easy to apply for it requires only connectivity indices in the calculations and does not require any experimental physicochemical properties in the calculation.