ABSTRACT
In the context of quantitative structure-activity relationship (QSAR) studies the traditional statistical indices, that is, simple correlation factor, standard error of estimates, as well as t-Student and Fisher tests are challenged against the algebraic spectral description of the multi-linear models. While the new correlation approach is built on the algebraic (Euclidian) norms and on the associate correlation factor, the validation and predictive features are assured throughout so called “ergodic” least action principle across the investigated models. The reliability of the present proposed spectral respecting statistical QSAR analysis is evaluated on modeling the aliphatic amines’ ecotoxicity by means of hydrophobicity, polarizability, and total energy structural parameters resulting in better predicting of molecular hierarchy of observed bioactivity being susceptible of as such general behavior.In QSAR studies, the “goodness of the fit” and “prediction ability” issues have been distinguished as the most important matters regarding the validation and usefulness of the provided models [1].
