QSPR Modeling of critical parameters of organic compounds belonging to different classes in terms of the simplex representation of molecular structure
Elena Mokshina, V. Е. Kuz’min, В. И. Недоступ · Russian Journal of Organic Chemistry · 2014
A set of QSPR models has been proposed for the description and prediction of critical temperatures, volumes, and pressures and Pitzer’s acentric factors for a large number of organic compounds on the basis of the simplex representation of molecular structure (SiRMS). The models have been designed with the aid of the nonlinear random forest learning algorithm. The SiRMS approach turned out to be an efficient tool for studying and predicting thermodynamic parameters of organic compounds. Taking into account the accuracy of the obtained results, obvious interpretability of simplex descriptors, and their effect on critical properties, the SiRMS approach may be recommended for studying and predicting critical properties.