Discovering general multidimensional associations

TL;DR

This paper introduces a new method for estimating the strength of multidimensional associations, which is more equitable, powerful, and scalable than existing measures like MIC, especially in higher dimensions and with covariates.

Contribution

The authors propose a generalized $R^2$ estimator for unknown relationships that outperforms MIC in power, convergence, and extends to multivariate and covariate-controlled scenarios.

Findings

Our approach is more equitable than MIC.

It has higher power to detect associations.

It converges faster with larger samples.

Abstract

When two variables are related by a known function, the coefficient of determination (denoted $R^2$) measures the proportion of the total variance in the observations that is explained by that function. This quantifies the strength of the relationship between variables by describing what proportion of the variance is signal as opposed to noise. For linear relationships, this is equal to the square of the correlation coefficient, $\rho$. When the parametric form of the relationship is unknown, however, it is unclear how to estimate the proportion of explained variance equitably - assigning similar values to equally noisy relationships. Here we demonstrate how to directly estimate a generalized $R^2$ when the form of the relationship is unknown, and we question the performance of the Maximal Information Coefficient (MIC) - a recently proposed information theoretic measure of dependence. We show that our approach behaves equitably, has more power than MIC to detect association between variables, and converges faster with increasing sample size. Most importantly, our approach generalizes to higher dimensions, which allows us to estimate the strength of multivariate relationships ($Y$ against $A,B, ...$) and to measure association while controlling for covariates ($Y$ against $X$ controlling for $C$).