Measuring statistical dependence between two random variables is a fundamental problem of data analysis, yet no single measure captures every type of relationship while remaining interpretable. Classical coefficients detect only narrow classes of dependence: linear for Pearson, monotonic for Spearman. More general measures are either computationally expensive, prone to overestimation, or uninformative about the structure of the relationship. This paper makes two contributions. First, we propose a unified framework for comparing dependence measures of different mathematical natures, based on the functional class of a coefficient and a set of linking properties anchored to the Pearson correlation as a common reference. Second, building on the correlation ratio, we develop a determination-based correlation coefficient (DBCC) that captures multivalued dependencies, in which a single value of the predictor may correspond to several states of the response. The coefficient decomposes the conditional distribution within local bins through a soft-assignment Bayesian Gaussian mixture model, replacing the discontinuous mode-seeking of an earlier kernel-based realisation. We prove the consistency of the estimator via an oracle decomposition that separates the algorithmic and statistical components of the error, and we show that the coefficient is computed in linear time. The construction retains the transparent variance-explained interpretation of the classical coefficient of determination $R^{2}$.

[1] Bickel P.J., Doksum K.A. Mathematical statistics: basic ideas and selected topics, Volume I, 2nd ed. Chapman and Hall/CRC, Boca Raton, 2015. doi:10.1201/b18312

[2] Bickel P.J., Freedman D.A. Some asymptotic theory for the bootstrap. Ann. Statist. 1981, 9 (6), 1196– 1217. doi:10.1214/aos/1176345637

[3] Chatterjee S. A new coefficient of correlation. J. Amer. Statist. Assoc. 2021, 116 (536), 2009–2022. doi:10.1080/01621459.2020.1758115

[4] Doran G. RDC: the randomized dependence coefficient in Python. Source code (software), 2014. https://github.com/garydoranjr/rdc (accessed June 2026).

[5] Gebelein H. Das statistische Problem der Korrelation als Variations- und Eigenwertproblem und sein Zusammenhang mit der Ausgleichsrechnung. Z. Angew. Math. Mech. 1941, 21 (6), 364–379. doi:10.1002/zamm.19410210604 (in German)

[6] Hirschfeld H.O. A connection between correlation and contingency. Math. Proc. Cambridge Philos. Soc. 1935, 31 (4), 520–524. doi:10.1017/S0305004100013517

[7] Lopez-Paz D., Hennig P., Sch¨olkopf B. The randomized dependence coefficient. In: Burges C.J.C., Bottou L., Welling M., Ghahramani Z., Weinberger K.Q. (Eds.) Advances in Neural Information Processing Systems 26 (NIPS 2013). Curran Associates, 2013, 1–9.

[8] McLachlan G.J., Peel D. Finite mixture models. In: Wiley Series in Probability and Statistics. John Wiley & Sons, New York, 2000. doi:10.1002/0471721182

[9] Pearson K. Note on regression and inheritance in the case of two parents. Proc. Roy. Soc. London 1895, 58, 240–242. doi:10.1098/rspl.1895.0041

[10] Pedregosa F., Varoquaux G., Gramfort A., Michel V., Thirion B., Grisel O., Blondel M., Prettenhofer P., Weiss R., Dubourg V., Vanderplas J., Passos A., Cournapeau D., Brucher M., Perrot M., Duchesnay E. Scikit-learn: machine learning in Python. J. Mach. Learn. Res. 2011, 12, 2825–2830.

[11] Ramos-Carre˜no C., Torrecilla J.L. dcor: distance correlation and energy statistics in Python. SoftwareX 2023, 22, 101326. doi:10.1016/j.softx.2023.101326

[12] Redner R.A., Walker H.F. Mixture densities, maximum likelihood and the EM algorithm. SIAM Rev. 1984, 26 (2), 195–239. doi:10.1137/1026034

[13] R´enyi A. On measures of dependence. Acta Math. Acad. Sci. Hungar. 1959, 10 (3–4), 441–451. doi:10.1007/BF02024507

[14] Reshef D.N., Reshef Y.A., Finucane H.K., Grossman S.R., McVean G., Turnbaugh P.J., Lander E.S., Mitzenmacher M., Sabeti P.C. Detecting novel associations in large data sets. Science 2011, 334 (6062), 1518–1524. doi:10.1126/science.1205438

[15] Rousseau J., Mengersen K. Asymptotic behaviour of the posterior distribution in overfitted mixture models. J. R. Stat. Soc. Ser. B Stat. Methodol. 2011, 73 (5), 689–710. doi:10.1111/j.1467-9868.2011.00781.x

[16] Scarsini M. On measures of concordance. Stochastica 1984, 8 (3), 201–218.

[17] Scott D.W. On optimal and data-based histograms. Biometrika 1979, 66 (3), 605–610. doi:10.1093/biomet/66.3.605

[18] Spearman C. The proof and measurement of association between two things. Amer. J. Psychol. 1904, 15 (1), 72–101. doi:10.2307/1412159

[19] Sz´ekely G.J., Rizzo M.L., Bakirov N.K. Measuring and testing dependence by correlation of distances. Ann. Statist. 2007, 35 (6), 2769–2794. doi:10.1214/009053607000000505

[20] Teicher H. Identifiability of finite mixtures. Ann. Math. Statist. 1963, 34 (4), 1265–1269. doi:10.1214/aoms/1177703862

[21] van der Vaart A.W. Asymptotic statistics. In: Cambridge Series in Statistical and Probabilistic Mathematics, 3. Cambridge University Press, Cambridge, 1998. doi:10.1017/CBO9780511802256

[22] van der Vaart A.W., Wellner J.A. Weak convergence and empirical processes. In: Springer Series in Statistics. Springer, New York, 1996. doi:10.1007/978-1-4757-2545-2

[23] Wang Y., Blei D.M. Frequentist consistency of variational Bayes. J. Amer. Statist. Assoc. 2019, 114 (527), 1147–1161. doi:10.1080/01621459.2018.1473776