Using a hybrid of the Genetic Algorithm (GA) and the Fmincon optimization techniques is proposed for reducing vapor liquid equilibrium (VLE) data of systems formed by a subcritical and a supercritical compound. As an example, this hybrid technique is used with advantage for evaluation of binary parameters required by cubic equations of state for the correlation of VLE data of binary mixtures containing carbon dioxide with 2-methoxyethanol and 2-ethoxyethanol. Four cubic equations of state (modified Van der Waals, Soave-Redlich-Kwong, Peng-Robinson, and Peng-Robinson-Stryjek-Vera), coupled with four mixing rules (conventional one-binary parameter, Margules-type, Huron-Vidal and the reformulated Wong-Sandler) were used. The Peng-Robinson-Stryjek-Vera EoS coupled with the Margules-type mixing rule gave the best correlation of vapor phase composition and combined with the reformulated Wong-Sandler mixing rule gave the best correlation of equilibrium pressure. The latter combination gave the best overall fit of the data. The binary interaction parameters were found to be independent of temperature. (c) 2012 Elsevier B.V. All rights reserved.