Nonparametric Measure of Similarity in Tasks of the Analysis of Network Structures

<p>At the heart of creation of network models of difficult structures the measure of similarity of behavior of elements of these structures lies. In the analysis of such structures according to casual data that is characteristic, for example, for tasks of the analysis of the stock markets, the standard measure of similarity of behavior of profitabilities of actions is the coefficient of correlation of Pearson. Apparently, wide popularity of this coefficient is the reason of it. At the same time, the selective analog of coefficient of correlation of Pearson which is rather steady and quite acceptable assessment of a measure of dependence of normally distributed random variables isn&#39;t steady at deviations from a normality. It results in need of research of other measures of dependence of the random variables describing, in particular, profitabilities of actions of stock markets.</p>