High-dimensional functional time series forecasting

Yuan Gao*, Hanlin Shang, Yanrong Yang

*Corresponding author for this work

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


In this paper, we address the problem of forecasting high-dimensional functional time series through a two-fold dimension reduction procedure. Dynamic functional principal component analysis is applied to reduce each infinite-dimension functional time series to a vector. We use factor model as a further dimension reduction technique so that only a small number of latent factors are preserved. Simple time series models can be used to forecast the factors and forecast of the functions can be constructed. The proposed method is easy to implement especially when the dimension of functional time series is large. We show the superiority of our approach by both simulation studies and an application to Japan mortality rates data.
Original languageEnglish
Title of host publicationFunctional statistics and related fields
EditorsGermán Aneiros, Enea G. Bongiorno, Ricardo Cao, Philippe Vieu
Place of PublicationCham
Number of pages6
ISBN (Electronic)9783319558462
ISBN (Print)9783319558455
Publication statusPublished - 2017
Externally publishedYes
EventInternational Workshop on Functional and Operatorial Statistics (4th : 2017) - La Coruña, Spain
Duration: 15 Jun 201717 Jun 2017

Publication series

NameContributions to statistics


ConferenceInternational Workshop on Functional and Operatorial Statistics (4th : 2017)
CityLa Coruña

Bibliographical note

Also published as: Gao, Y., Shang, H. L., & Yang, Y. (2019). High-dimensional functional time series forecasting: An application to age-specific mortality rates. Journal of Multivariate Analysis, 170, 232-243. DOI: 10.1016/j.jmva.2018.10.003


  • Time Series Model
  • Principal Component Score
  • Multiple Time Series
  • Functional Principal Component
  • Dimension Reduction Technique


Dive into the research topics of 'High-dimensional functional time series forecasting'. Together they form a unique fingerprint.

Cite this