On a nonparametric notion of residual and its applications

DOI10.1016/J.SPL.2015.10.011zbMATH Open1383.62169arXiv1409.3886OpenAlexW2162293148MaRDI QIDQ899668FDOQ899668

Rohit Kumar Patra, G. J. Székely, Bodhisattva Sen

Publication date: 30 December 2015

Published in: Statistics \& Probability Letters (Search for Journal in Brave)

Abstract: Let $(X,mathbfZ)$ be a continuous random vector in $mathbbRimesmathbb{R}^d$, $dge1$. In this paper, we define the notion of a nonparametric residual of $X$ on $mathbfZ$ that is always independent of the predictor $mathbfZ$. We study its properties and show that the proposed notion of residual matches with the usual residual (error) in a multivariate normal regression model. Given a random vector $(X,Y,mathbfZ)$ in $mathbbRimesmathbbRimesmathbb{R}^d$, we use this notion of residual to show that the conditional independence between $X$ and $Y$, given $mathbfZ$, is equivalent to the mutual independence of the residuals (of $X$ on $mathbfZ$ and $Y$ on $mathbfZ$) and $mathbfZ$. This result is used to develop a test for conditional independence. We propose a bootstrap scheme to approximate the critical value of this test. We compare the proposed test, which is easily implementable, with some of the existing procedures through a simulation study.

Full work available at URL: https://arxiv.org/abs/1409.3886

zbMATH Keywords

conditional distribution functiontesting conditional independence

Mathematics Subject Classification ID

Nonparametric hypothesis testing (62G10) Hypothesis testing in multivariate analysis (62H15) Measures of association (correlation, canonical correlation, etc.) (62H20)

Cites Work

• Measuring and testing dependence by correlation of distances
• Partial distance correlation with methods for dissimilarities
• Title not available (Why is that?)
• Algorithmic Learning Theory
• A new test for multivariate normality
• Brownian distance covariance
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Testing independence and goodness-of-fit in linear models
• A NONPARAMETRIC HELLINGER METRIC TEST FOR CONDITIONAL INDEPENDENCE
• Energy statistics: a class of statistics based on distances
• Testing conditional independence using maximal nonlinear conditional correlation
• Nonparametric multivariate descriptive measures based on spatial quantiles
• Title not available (Why is that?)
• Strongly consistent nonparametric tests of conditional independence
• On Fractile Transformation of Covariates in Regression

Cited In (10)

• Measures of conditional dependence for nonlinearity, asymmetry and beyond
• Minimax optimal conditional independence testing
• Local permutation tests for conditional independence
• Distance-covariance-based tests for heteroscedasticity in nonlinear regressions
• The Locally Gaussian Partial Correlation
• A survey of some recent developments in measures of association
• Title not available (Why is that?)
• Characterizations of residual implications derived from uni-nullnorms
• A simple measure of conditional dependence
• General tests of conditional independence based on empirical processes indexed by functions

Uses Software

• R
• energy

This page was built for publication: On a nonparametric notion of residual and its applications

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q899668)