ANALYSIS OF THE REACTIVITY OF SINGLE BONDS IN ALIPHATIC MOLECULES BY STATISTICAL AND PATTERN-RECOGNITION METHODS

The polar breaking of a bond is the initiating step in many organic re actions. The factors influencing the heterolysis of a bond are investi gated by a series of statistical and pattern recognition methods. The power of a variety of methods-including principal component analysis, linear discriminant analysis, cluster analysis, k-nearest-neighbor ana lysis, and logistic regression analysis-for understanding and predicti ng the reactivity of bonds based on electronic and energy parameters i s elaborated. In particular, we show how information on whether a seri es of bonds will undergo heterolysis or not can be used to derive a fu nction quantifying polar reactivity. In the end, an equation is develo ped that allows the prediction of the polar breaking of bonds for a wi de range of aliphatic structures.