Modelling functional data with high-dimensional error structure

Yuan Gao*, Hanlin Shang, Yanrong Yang

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference proceeding contributionpeer-review

Abstract

We propose to model rawfunctional data as a mixture of functions and highdimensional error. The conventional approach to retrieve the functional component from raw data is through varied smoothing techniques. Nevertheless, smoothing itself may not be adequate when measurement error exists.We propose to use factor model to reduce the dimension of the high-dimensional measurement error, while smoothing the functional component. Our model also provides as an alternative for modelling functional data with step jump. Regularized least squares method is used to find the model estimates. We look at the asymptotic behaviour of the estimator when both the sample size and the number of points per curve go to infinity and the limiting distribution is derived.
Original languageEnglish
Title of host publicationFunctional and high-dimensional statistics and related fields
EditorsGermán Aneiros, Ivana Horová, Marie Huésková, Philippe Vieu
Place of PublicationCham
PublisherSpringer, Springer Nature
Pages99-106
Number of pages8
ISBN (Electronic)9783030477561
ISBN (Print)9783030477554
DOIs
Publication statusPublished - Jun 2020
EventInternational Workshop on Functional and Operatorial Statistics (5th : 2020) - Brno, Czech Republic
Duration: 24 Jun 202026 Jun 2020

Publication series

NameContributions to Statistics
PublisherSpringer
ISSN (Print)1431-1968

Conference

ConferenceInternational Workshop on Functional and Operatorial Statistics (5th : 2020)
Country/TerritoryCzech Republic
CityBrno
Period24/06/2026/06/20

Fingerprint

Dive into the research topics of 'Modelling functional data with high-dimensional error structure'. Together they form a unique fingerprint.

Cite this