High precision correlations of thermo-physical properties

High precision correlation of physical and thermodynamic properties is considered. A new general type of correlation is proposed, where the correlations contain one or more theory-based terms, which are augmented by polynomial terms in order to compensate for non-idealities in the data that are not explained by theory. An extended version of the SROV program of Shacham and Brauner (1999, 2003) is used to select the most influential terms that should be included in the correlation, to detect outlying measurements and to select the optimal parameter for Box-Cox (maximum likelihood) transformations. Examples are presented, where the application of new algorithm enabled obtaining stable correlations with a precision of several orders of magnitude higher than that of published correlations.