Reza Sharafdini

A comparative study based on the structure-property regression analysis is performed in order to test and evaluate the application possibilities of various graph irregularity indices for the prediction of physico-chemical properties of octane isomers. By restricting attention to single-variable linear regressions, we investigate the stochastic relationships between 18 preselected irregularity indices and 5 physico-chemical properties of octane isomers. These are: Boiling point (Bp), Standard enthalpy of vaporization (DHVAP), Acentric factor (AcenFac), Enthalpy of vaporization (HVAP) and Entropy. The degree of the intercorrelation was evaluated by traditional correlation coefficients. In physico-chemical applications, it is a widely accepted but theoretically not verified belief is that the use of graph irregularity indices are not to be efficient in QSPR studies of molecular graphs. Our observations refute this preconception. Presenting demonstrative counter-examples it is shown that there exist several irregularity indices by which four octane isomer properties (DHVAP, Entropy, AcenFac and HVAP) can be predicted with a good accuracy.