Mean vector testing for high-dimensional dependent observations

Deepak Nag Ayyala, Junyong Park, Anindya Roy

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

When testing for the mean vector in a high-dimensional setting, it is generally assumed that the observations are independently and identically distributed. However if the data are dependent, the existing test procedures fail to preserve type I error at a given nominal significance level. We propose a new test for the mean vector when the dimension increases linearly with sample size and the data is a realization of an M-dependent stationary process. The order M is also allowed to increase with the sample size. Asymptotic normality of the test statistic is derived by extending the Central Limit Theorem for M-dependent processes using two-dimensional triangular arrays. The cost of ignoring dependence among observations is assessed in finite samples through simulations.

Original languageEnglish (US)
Pages (from-to)136-155
Number of pages20
JournalJournal of Multivariate Analysis
Volume153
DOIs
StatePublished - Jan 1 2017

Fingerprint

Dependent Observations
High-dimensional
Testing
Dependent
Sample Size
Triangular Array
Significance level
Type I error
Statistics
Stationary Process
Asymptotic Normality
Central limit theorem
Identically distributed
Test Statistic
Categorical or nominal
Linearly
Costs
Sample size
Simulation
Observation

Keywords

  • Asymptotic normality
  • Dependent data
  • High-dimension
  • Mean vector testing
  • Triangular array

ASJC Scopus subject areas

  • Statistics and Probability
  • Numerical Analysis
  • Statistics, Probability and Uncertainty

Cite this

Mean vector testing for high-dimensional dependent observations. / Ayyala, Deepak Nag; Park, Junyong; Roy, Anindya.

In: Journal of Multivariate Analysis, Vol. 153, 01.01.2017, p. 136-155.

Research output: Contribution to journalArticle

@article{4cef77cf7b8c4dcc8c90007281ad043e,
title = "Mean vector testing for high-dimensional dependent observations",
abstract = "When testing for the mean vector in a high-dimensional setting, it is generally assumed that the observations are independently and identically distributed. However if the data are dependent, the existing test procedures fail to preserve type I error at a given nominal significance level. We propose a new test for the mean vector when the dimension increases linearly with sample size and the data is a realization of an M-dependent stationary process. The order M is also allowed to increase with the sample size. Asymptotic normality of the test statistic is derived by extending the Central Limit Theorem for M-dependent processes using two-dimensional triangular arrays. The cost of ignoring dependence among observations is assessed in finite samples through simulations.",
keywords = "Asymptotic normality, Dependent data, High-dimension, Mean vector testing, Triangular array",
author = "Ayyala, {Deepak Nag} and Junyong Park and Anindya Roy",
year = "2017",
month = "1",
day = "1",
doi = "10.1016/j.jmva.2016.09.012",
language = "English (US)",
volume = "153",
pages = "136--155",
journal = "Journal of Multivariate Analysis",
issn = "0047-259X",
publisher = "Academic Press Inc.",

}

TY - JOUR

T1 - Mean vector testing for high-dimensional dependent observations

AU - Ayyala, Deepak Nag

AU - Park, Junyong

AU - Roy, Anindya

PY - 2017/1/1

Y1 - 2017/1/1

N2 - When testing for the mean vector in a high-dimensional setting, it is generally assumed that the observations are independently and identically distributed. However if the data are dependent, the existing test procedures fail to preserve type I error at a given nominal significance level. We propose a new test for the mean vector when the dimension increases linearly with sample size and the data is a realization of an M-dependent stationary process. The order M is also allowed to increase with the sample size. Asymptotic normality of the test statistic is derived by extending the Central Limit Theorem for M-dependent processes using two-dimensional triangular arrays. The cost of ignoring dependence among observations is assessed in finite samples through simulations.

AB - When testing for the mean vector in a high-dimensional setting, it is generally assumed that the observations are independently and identically distributed. However if the data are dependent, the existing test procedures fail to preserve type I error at a given nominal significance level. We propose a new test for the mean vector when the dimension increases linearly with sample size and the data is a realization of an M-dependent stationary process. The order M is also allowed to increase with the sample size. Asymptotic normality of the test statistic is derived by extending the Central Limit Theorem for M-dependent processes using two-dimensional triangular arrays. The cost of ignoring dependence among observations is assessed in finite samples through simulations.

KW - Asymptotic normality

KW - Dependent data

KW - High-dimension

KW - Mean vector testing

KW - Triangular array

UR - http://www.scopus.com/inward/record.url?scp=84992144158&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84992144158&partnerID=8YFLogxK

U2 - 10.1016/j.jmva.2016.09.012

DO - 10.1016/j.jmva.2016.09.012

M3 - Article

VL - 153

SP - 136

EP - 155

JO - Journal of Multivariate Analysis

JF - Journal of Multivariate Analysis

SN - 0047-259X

ER -