Wrażliwość miar dywergencji jako mierników niepodobieństwa struktur
Sensitivity of divergence measures as structure dissimilarity measurements

Summary/Abstract: The analyses of social and economic phenomena often involve the issue of similarity between business objects characterized by structure indicators. Usually, measures used for quantifying similarity or the lack of similarity between structures are a function of the distance metrics of their partial indicators. An examination of the similarity between structures can also apply divergence measures. This article indicates the possibility of using Csiszár class divergence measures (f-divergences), in particular: Hellinger discrimination, triangular discrimination, symmetric Chi-square divergence, arithmetic-geometric mean divergence, Kullback-Leibler divergence and Jensen-Shannon divergence to evaluate the degree of discrepancy between structures. The aim of the article is to examine the sensitivity of the indicated measures to the changes in the degree of discrepancy between structures.

• Details
• Contents