Key points are not available for this paper at this time.
Quantitative structure-property relationship (QSPR) modeling employs contemporary mathematical tools to predict physicochemical, thermodynamic and biological properties of compounds from their chemical structures. Graphical indices provide those mathematical tools and efficiently correlate physicochemical, thermodynamic and biological properties of compounds. This paper considers the class of benzenoid hydrocarbons (BHs) and investigates the predictive power of commonly occurring graphical temperature indices for determining thermodynamic characteristics of BHs. The entropy (So) & heat capacity (Cp) have been selected to advocate for thermodynamic characteristics. In order to validate the statistical inference, the lower 30 BHs have been opted as test molecules as their experimental data is also publicly available. First, a computer-based method is put forwarded to evaluate temperature indices of a chemical graph. The computational method is, then, employed to compute temperature indices of 30 lower BHs. Certain QPSR models are proposed by utilizing the experimental data of Cp and So for the BHs. Our statistical analysis suggests that the most efficient regression models are, in fact, quadratic. Unexpectedly, some underresearched temperature indices like the general first & second temperature indices perform well having the correlation coefficient >0.9 which is the minimum threshold in a comparative testing. Based on our statistical analysis, three recently proposed temperature indices perform exceptionally well in comparison with all the existing temperature indices. The results suggest their further employability in QSPR modeling. Importantly, our research contributes towards countering proliferation of graphical indices.
Hayat et al. (Tue,) studied this question.