A new measure of dependence: Integrated $R^2$

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Azadkia, Mona, Roudaki, Pouya
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915724824412160
author Azadkia, Mona
Roudaki, Pouya
author_facet Azadkia, Mona
Roudaki, Pouya
contents We introduce a novel measure of dependence that captures the extent to which a random variable $Y$ is determined by a random vector $X$. The measure equals zero precisely when $Y$ and $X$ are independent, and it attains one exactly when $Y$ is almost surely a measurable function of $X$. We further extend this framework to define a measure of conditional dependence between $Y$ and $X$ given $Z$. We propose a simple and interpretable estimator with computational complexity comparable to classical correlation coefficients, including those of Pearson, Spearman, and Chatterjee. Leveraging this dependence measure, we develop a tuning-free, model-agnostic variable selection procedure and establish its consistency under appropriate sparsity conditions. Extensive experiments on synthetic and real datasets highlight the strong empirical performance of our methodology and demonstrate substantial gains over existing approaches.
format Preprint
id arxiv_https___arxiv_org_abs_2505_18146
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A new measure of dependence: Integrated $R^2$
Azadkia, Mona
Roudaki, Pouya
Statistics Theory
Information Theory
Probability
Methodology
62H20, 62H15
We introduce a novel measure of dependence that captures the extent to which a random variable $Y$ is determined by a random vector $X$. The measure equals zero precisely when $Y$ and $X$ are independent, and it attains one exactly when $Y$ is almost surely a measurable function of $X$. We further extend this framework to define a measure of conditional dependence between $Y$ and $X$ given $Z$. We propose a simple and interpretable estimator with computational complexity comparable to classical correlation coefficients, including those of Pearson, Spearman, and Chatterjee. Leveraging this dependence measure, we develop a tuning-free, model-agnostic variable selection procedure and establish its consistency under appropriate sparsity conditions. Extensive experiments on synthetic and real datasets highlight the strong empirical performance of our methodology and demonstrate substantial gains over existing approaches.
title A new measure of dependence: Integrated $R^2$
topic Statistics Theory
Information Theory
Probability
Methodology
62H20, 62H15
url https://arxiv.org/abs/2505.18146