On Confidence Estimation Based on Quantitative Similarity Coefficients
I. V. Rodionov, А. Н. Созонтов · Automation and Remote Control · 2020
We consider the problem of estimating the accuracy of quantitative similarity coefficients. For this purpose, we introduce a new concept of the similarity measure for the corresponding coefficient. We show that only frequency forms of quantitative similarity coefficients represent consistent estimates of their similarity measures. We obtain asymptotic confidence intervals for the RužiŽka and Bray-Curtis similarity measures based on the coefficients with the same names. We also propose a test for homogeneity of two populations based on the above-mentioned coefficients.