Partial martingale difference correlation

Research output: Contribution to journalArticle

Abstract

We introduce the partial martingale difference correlation, a scalar-valued measure of conditional mean dependence of Y given X, adjusting for the nonlinear dependence on Z, where X, Y and Z are random vectors of arbitrary dimensions. At the population level, partial martingale difference correlation is a natural extension of partial distance correlation developed recently by Székely and Rizzo [14], which characterizes the dependence of Y and X, after controlling for the nonlinear effect of Z. It extends the martingale difference correlation first introduced in Shao and Zhang [10] just as partial distance correlation extends the distance correlation in Székely, Rizzo and Bakirov [13]. Sample partial martingale difference correlation is also defined building on some new results on equivalent expressions of sample martingale difference correlation. Numerical results demonstrate the effectiveness of these new dependence measures in the context of variable selection and dependence testing.

Original languageEnglish (US)
Pages (from-to)1492-1517
Number of pages26
JournalElectronic Journal of Statistics
Volume9
DOIs
StatePublished - Jan 1 2015

Fingerprint

Martingale Difference
Partial
Selection of Variables
Martingale
Nonlinear Effects
Natural Extension
Random Vector
Scalar
Numerical Results
Testing

Keywords

  • Distance correlation
  • Nonlinear dependence
  • Partial correlation
  • Variable selection

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Partial martingale difference correlation. / Park, Trevor; Shao, Xiaofeng; Yao, Shun.

In: Electronic Journal of Statistics, Vol. 9, 01.01.2015, p. 1492-1517.

Research output: Contribution to journalArticle

@article{797569530f2e46189f26cd6773487a59,
title = "Partial martingale difference correlation",
abstract = "We introduce the partial martingale difference correlation, a scalar-valued measure of conditional mean dependence of Y given X, adjusting for the nonlinear dependence on Z, where X, Y and Z are random vectors of arbitrary dimensions. At the population level, partial martingale difference correlation is a natural extension of partial distance correlation developed recently by Sz{\'e}kely and Rizzo [14], which characterizes the dependence of Y and X, after controlling for the nonlinear effect of Z. It extends the martingale difference correlation first introduced in Shao and Zhang [10] just as partial distance correlation extends the distance correlation in Sz{\'e}kely, Rizzo and Bakirov [13]. Sample partial martingale difference correlation is also defined building on some new results on equivalent expressions of sample martingale difference correlation. Numerical results demonstrate the effectiveness of these new dependence measures in the context of variable selection and dependence testing.",
keywords = "Distance correlation, Nonlinear dependence, Partial correlation, Variable selection",
author = "Trevor Park and Xiaofeng Shao and Shun Yao",
year = "2015",
month = "1",
day = "1",
doi = "10.1214/15-EJS1047",
language = "English (US)",
volume = "9",
pages = "1492--1517",
journal = "Electronic Journal of Statistics",
issn = "1935-7524",
publisher = "Institute of Mathematical Statistics",

}

TY - JOUR

T1 - Partial martingale difference correlation

AU - Park, Trevor

AU - Shao, Xiaofeng

AU - Yao, Shun

PY - 2015/1/1

Y1 - 2015/1/1

N2 - We introduce the partial martingale difference correlation, a scalar-valued measure of conditional mean dependence of Y given X, adjusting for the nonlinear dependence on Z, where X, Y and Z are random vectors of arbitrary dimensions. At the population level, partial martingale difference correlation is a natural extension of partial distance correlation developed recently by Székely and Rizzo [14], which characterizes the dependence of Y and X, after controlling for the nonlinear effect of Z. It extends the martingale difference correlation first introduced in Shao and Zhang [10] just as partial distance correlation extends the distance correlation in Székely, Rizzo and Bakirov [13]. Sample partial martingale difference correlation is also defined building on some new results on equivalent expressions of sample martingale difference correlation. Numerical results demonstrate the effectiveness of these new dependence measures in the context of variable selection and dependence testing.

AB - We introduce the partial martingale difference correlation, a scalar-valued measure of conditional mean dependence of Y given X, adjusting for the nonlinear dependence on Z, where X, Y and Z are random vectors of arbitrary dimensions. At the population level, partial martingale difference correlation is a natural extension of partial distance correlation developed recently by Székely and Rizzo [14], which characterizes the dependence of Y and X, after controlling for the nonlinear effect of Z. It extends the martingale difference correlation first introduced in Shao and Zhang [10] just as partial distance correlation extends the distance correlation in Székely, Rizzo and Bakirov [13]. Sample partial martingale difference correlation is also defined building on some new results on equivalent expressions of sample martingale difference correlation. Numerical results demonstrate the effectiveness of these new dependence measures in the context of variable selection and dependence testing.

KW - Distance correlation

KW - Nonlinear dependence

KW - Partial correlation

KW - Variable selection

UR - http://www.scopus.com/inward/record.url?scp=84937856954&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84937856954&partnerID=8YFLogxK

U2 - 10.1214/15-EJS1047

DO - 10.1214/15-EJS1047

M3 - Article

AN - SCOPUS:84937856954

VL - 9

SP - 1492

EP - 1517

JO - Electronic Journal of Statistics

JF - Electronic Journal of Statistics

SN - 1935-7524

ER -