Structured functional principal component analysis

Haochang Shou, Vadim Zipunnikov, Ciprian M Crainiceanu, Sonja Greven

Research output: Contribution to journalArticle

Abstract

Motivated by modern observational studies, we introduce a class of functional models that expand nested and crossed designs. These models account for the natural inheritance of the correlation structures from sampling designs in studies where the fundamental unit is a function or image. Inference is based on functional quadratics and their relationship with the underlying covariance structure of the latent processes. A computationally fast and scalable estimation procedure is developed for high-dimensional data. Methods are used in applications including high-frequency accelerometer data for daily activity, pitch linguistic data for phonetic analysis, and EEG data for studying electrical brain activity during sleep.

Original languageEnglish (US)
Pages (from-to)247-257
Number of pages11
JournalBiometrics
Volume71
Issue number1
DOIs
StatePublished - Mar 1 2015

Fingerprint

Functional Principal Component Analysis
Principal Component Analysis
Principal component analysis
principal component analysis
Phonetics
Speech analysis
observational studies
Linguistics
Electroencephalography
sleep
Accelerometers
Latent Process
Fundamental Units
Observational Studies
Quadratic Functional
Brain
data analysis
Sleep
Observational Study
Functional Model

Keywords

  • Functional linear mixed model
  • Functional principal component analysis
  • Latent process
  • Multilevel correlation structure
  • Variance component

ASJC Scopus subject areas

  • Statistics and Probability
  • Medicine(all)
  • Immunology and Microbiology(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Agricultural and Biological Sciences(all)
  • Applied Mathematics

Cite this

Structured functional principal component analysis. / Shou, Haochang; Zipunnikov, Vadim; Crainiceanu, Ciprian M; Greven, Sonja.

In: Biometrics, Vol. 71, No. 1, 01.03.2015, p. 247-257.

Research output: Contribution to journalArticle

Shou, Haochang ; Zipunnikov, Vadim ; Crainiceanu, Ciprian M ; Greven, Sonja. / Structured functional principal component analysis. In: Biometrics. 2015 ; Vol. 71, No. 1. pp. 247-257.
@article{9ef7a85803d84f57a0bb77ce48853444,
title = "Structured functional principal component analysis",
abstract = "Motivated by modern observational studies, we introduce a class of functional models that expand nested and crossed designs. These models account for the natural inheritance of the correlation structures from sampling designs in studies where the fundamental unit is a function or image. Inference is based on functional quadratics and their relationship with the underlying covariance structure of the latent processes. A computationally fast and scalable estimation procedure is developed for high-dimensional data. Methods are used in applications including high-frequency accelerometer data for daily activity, pitch linguistic data for phonetic analysis, and EEG data for studying electrical brain activity during sleep.",
keywords = "Functional linear mixed model, Functional principal component analysis, Latent process, Multilevel correlation structure, Variance component",
author = "Haochang Shou and Vadim Zipunnikov and Crainiceanu, {Ciprian M} and Sonja Greven",
year = "2015",
month = "3",
day = "1",
doi = "10.1111/biom.12236",
language = "English (US)",
volume = "71",
pages = "247--257",
journal = "Biometrics",
issn = "0006-341X",
publisher = "Wiley-Blackwell",
number = "1",

}

TY - JOUR

T1 - Structured functional principal component analysis

AU - Shou, Haochang

AU - Zipunnikov, Vadim

AU - Crainiceanu, Ciprian M

AU - Greven, Sonja

PY - 2015/3/1

Y1 - 2015/3/1

N2 - Motivated by modern observational studies, we introduce a class of functional models that expand nested and crossed designs. These models account for the natural inheritance of the correlation structures from sampling designs in studies where the fundamental unit is a function or image. Inference is based on functional quadratics and their relationship with the underlying covariance structure of the latent processes. A computationally fast and scalable estimation procedure is developed for high-dimensional data. Methods are used in applications including high-frequency accelerometer data for daily activity, pitch linguistic data for phonetic analysis, and EEG data for studying electrical brain activity during sleep.

AB - Motivated by modern observational studies, we introduce a class of functional models that expand nested and crossed designs. These models account for the natural inheritance of the correlation structures from sampling designs in studies where the fundamental unit is a function or image. Inference is based on functional quadratics and their relationship with the underlying covariance structure of the latent processes. A computationally fast and scalable estimation procedure is developed for high-dimensional data. Methods are used in applications including high-frequency accelerometer data for daily activity, pitch linguistic data for phonetic analysis, and EEG data for studying electrical brain activity during sleep.

KW - Functional linear mixed model

KW - Functional principal component analysis

KW - Latent process

KW - Multilevel correlation structure

KW - Variance component

UR - http://www.scopus.com/inward/record.url?scp=84961291511&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84961291511&partnerID=8YFLogxK

U2 - 10.1111/biom.12236

DO - 10.1111/biom.12236

M3 - Article

C2 - 25327216

AN - SCOPUS:84961291511

VL - 71

SP - 247

EP - 257

JO - Biometrics

JF - Biometrics

SN - 0006-341X

IS - 1

ER -