# Dynamic functional principal components

Siegfried Hörmann, Lukasz Kidziski, Marc Hallin

Research output: Contribution to journalArticleResearchpeer-review

### Abstract

We address the problem of dimension reduction for time series of functional data $(X_t : t\in Z)$. Such functional time series frequently arise, for example, when a continuous time process is segmented into some smaller natural units, such as days. Then each Xt represents one intraday curve. We argue that functional principal component analysis, though a key technique in the field and a benchmark for any competitor, does not provide an adequate dimension reduction in a time series setting. Functional principal component analysis indeed is a static procedure which ignores the essential information that is provided by the serial dependence structure of the functional data under study. Therefore, inspired by Brillinger's theory of dynamic principal components, we propose a dynamic version of functional principal component analysis which is based on a frequency domain approach. By means of a simulation study and an empirical illustration, we show the considerable improvement that the dynamic approach entails when compared with the usual static procedure.
Language English 319-348 30 Journal of the Royal Statistical Society / B 77 2 Published - 2015

### Fingerprint

Functional Principal Component Analysis
Principal Components
Functional Data
Time series
Dimension Reduction
Serial Dependence
Dependence Structure
Frequency Domain
Continuous Time
Simulation Study
Benchmark
Curve
Unit
Principal components
Principal component analysis
Dimension reduction

### Cite this

Dynamic functional principal components. / Hörmann, Siegfried; Kidziski, Lukasz; Hallin, Marc.

In: Journal of the Royal Statistical Society / B, Vol. 77, No. 2, 2015, p. 319-348.

Research output: Contribution to journalArticleResearchpeer-review

Hörmann, S, Kidziski, L & Hallin, M 2015, 'Dynamic functional principal components' Journal of the Royal Statistical Society / B, vol. 77, no. 2, pp. 319-348.
Hörmann, Siegfried ; Kidziski, Lukasz ; Hallin, Marc. / Dynamic functional principal components. In: Journal of the Royal Statistical Society / B. 2015 ; Vol. 77, No. 2. pp. 319-348
@article{522bb555a497480191b2bf9bcc9c6ee3,
title = "Dynamic functional principal components",
abstract = "We address the problem of dimension reduction for time series of functional data $(X_t : t\in Z)$. Such functional time series frequently arise, for example, when a continuous time process is segmented into some smaller natural units, such as days. Then each Xt represents one intraday curve. We argue that functional principal component analysis, though a key technique in the field and a benchmark for any competitor, does not provide an adequate dimension reduction in a time series setting. Functional principal component analysis indeed is a static procedure which ignores the essential information that is provided by the serial dependence structure of the functional data under study. Therefore, inspired by Brillinger's theory of dynamic principal components, we propose a dynamic version of functional principal component analysis which is based on a frequency domain approach. By means of a simulation study and an empirical illustration, we show the considerable improvement that the dynamic approach entails when compared with the usual static procedure.",
author = "Siegfried H{\"o}rmann and Lukasz Kidziski and Marc Hallin",
year = "2015",
language = "English",
volume = "77",
pages = "319--348",
journal = "Journal of the Royal Statistical Society / B",
issn = "1369-7412",
publisher = "Wiley-Blackwell",
number = "2",

}

TY - JOUR

T1 - Dynamic functional principal components

AU - Hörmann,Siegfried

AU - Kidziski,Lukasz

AU - Hallin,Marc

PY - 2015

Y1 - 2015

N2 - We address the problem of dimension reduction for time series of functional data $(X_t : t\in Z)$. Such functional time series frequently arise, for example, when a continuous time process is segmented into some smaller natural units, such as days. Then each Xt represents one intraday curve. We argue that functional principal component analysis, though a key technique in the field and a benchmark for any competitor, does not provide an adequate dimension reduction in a time series setting. Functional principal component analysis indeed is a static procedure which ignores the essential information that is provided by the serial dependence structure of the functional data under study. Therefore, inspired by Brillinger's theory of dynamic principal components, we propose a dynamic version of functional principal component analysis which is based on a frequency domain approach. By means of a simulation study and an empirical illustration, we show the considerable improvement that the dynamic approach entails when compared with the usual static procedure.

AB - We address the problem of dimension reduction for time series of functional data $(X_t : t\in Z)$. Such functional time series frequently arise, for example, when a continuous time process is segmented into some smaller natural units, such as days. Then each Xt represents one intraday curve. We argue that functional principal component analysis, though a key technique in the field and a benchmark for any competitor, does not provide an adequate dimension reduction in a time series setting. Functional principal component analysis indeed is a static procedure which ignores the essential information that is provided by the serial dependence structure of the functional data under study. Therefore, inspired by Brillinger's theory of dynamic principal components, we propose a dynamic version of functional principal component analysis which is based on a frequency domain approach. By means of a simulation study and an empirical illustration, we show the considerable improvement that the dynamic approach entails when compared with the usual static procedure.

M3 - Article

VL - 77

SP - 319

EP - 348

JO - Journal of the Royal Statistical Society / B

T2 - Journal of the Royal Statistical Society / B

JF - Journal of the Royal Statistical Society / B

SN - 1369-7412

IS - 2

ER -