Gaussian process methods for nonparametric functional regression with mixed predictors

Bo Wang, Aiping Xu

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Gaussian process methods are proposed for nonparametric functional regression for both scalar and functional responses with mixed multidimensional functional and scalar predictors. The proposed models allow the response variables to depend on the entire trajectories of the functional predictors. They inherit the desirable properties of Gaussian process regression, and can naturally accommodate both scalar and functional variables as the predictors, as well as easy to obtain and express uncertainty in predictions. The numerical experiments show that the proposed methods significantly outperform the competing models, and their usefulness is also demonstrated by the application to two real datasets.
Original languageEnglish
Pages (from-to)80-90
Number of pages11
JournalComputational Statistics and Data Analysis
Volume131
Early online date26 Jul 2018
DOIs
Publication statusPublished - Mar 2019

Fingerprint

Gaussian Process
Predictors
Regression
Scalar
Functional Response
Express
Numerical Experiment
Entire
Trajectory
Uncertainty
Prediction
Model

Keywords

  • Functional principal component analysis
  • Functional regression
  • Gaussian process regression
  • Nonparametric methods
  • Semi-metric

ASJC Scopus subject areas

  • Statistics and Probability
  • Computational Mathematics
  • Computational Theory and Mathematics
  • Applied Mathematics

Cite this

Gaussian process methods for nonparametric functional regression with mixed predictors. / Wang, Bo; Xu, Aiping.

In: Computational Statistics and Data Analysis, Vol. 131, 03.2019, p. 80-90.

Research output: Contribution to journalArticle

@article{5a965b12196f4c46a2cf425a6d974064,
title = "Gaussian process methods for nonparametric functional regression with mixed predictors",
abstract = "Gaussian process methods are proposed for nonparametric functional regression for both scalar and functional responses with mixed multidimensional functional and scalar predictors. The proposed models allow the response variables to depend on the entire trajectories of the functional predictors. They inherit the desirable properties of Gaussian process regression, and can naturally accommodate both scalar and functional variables as the predictors, as well as easy to obtain and express uncertainty in predictions. The numerical experiments show that the proposed methods significantly outperform the competing models, and their usefulness is also demonstrated by the application to two real datasets.",
keywords = "Functional principal component analysis, Functional regression, Gaussian process regression, Nonparametric methods, Semi-metric",
author = "Bo Wang and Aiping Xu",
year = "2019",
month = "3",
doi = "10.1016/j.csda.2018.07.009",
language = "English",
volume = "131",
pages = "80--90",
journal = "Computational Statistics and Data Analysis",
issn = "0167-9473",
publisher = "Elsevier",

}

TY - JOUR

T1 - Gaussian process methods for nonparametric functional regression with mixed predictors

AU - Wang, Bo

AU - Xu, Aiping

PY - 2019/3

Y1 - 2019/3

N2 - Gaussian process methods are proposed for nonparametric functional regression for both scalar and functional responses with mixed multidimensional functional and scalar predictors. The proposed models allow the response variables to depend on the entire trajectories of the functional predictors. They inherit the desirable properties of Gaussian process regression, and can naturally accommodate both scalar and functional variables as the predictors, as well as easy to obtain and express uncertainty in predictions. The numerical experiments show that the proposed methods significantly outperform the competing models, and their usefulness is also demonstrated by the application to two real datasets.

AB - Gaussian process methods are proposed for nonparametric functional regression for both scalar and functional responses with mixed multidimensional functional and scalar predictors. The proposed models allow the response variables to depend on the entire trajectories of the functional predictors. They inherit the desirable properties of Gaussian process regression, and can naturally accommodate both scalar and functional variables as the predictors, as well as easy to obtain and express uncertainty in predictions. The numerical experiments show that the proposed methods significantly outperform the competing models, and their usefulness is also demonstrated by the application to two real datasets.

KW - Functional principal component analysis

KW - Functional regression

KW - Gaussian process regression

KW - Nonparametric methods

KW - Semi-metric

UR - http://www.scopus.com/inward/record.url?scp=85050944442&partnerID=8YFLogxK

U2 - 10.1016/j.csda.2018.07.009

DO - 10.1016/j.csda.2018.07.009

M3 - Article

VL - 131

SP - 80

EP - 90

JO - Computational Statistics and Data Analysis

JF - Computational Statistics and Data Analysis

SN - 0167-9473

ER -