Geostatistical model averaging based on conditional information criteria

Chun-Shu Chen, Hsin Cheng Huang

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

Variable selection in geostatistical regression is an important problem, but has not been well studied in the literature. In this paper, we focus on spatial prediction and consider a class of conditional information criteria indexed by a penalty parameter. Instead of applying a fixed criterion, which leads to an unstable predictor in the sense that it is discontinuous with respect to the response variables due to that a small change in the response may cause a different model to be selected, we further stabilize the predictor by local model averaging, resulting in a predictor that is not only continuous but also differentiable even after plugging-in estimated model parameters. Then Stein's unbiased risk estimate is applied to select the penalty parameter, leading to a data-dependent penalty that is adaptive to the underlying model. Some numerical experiments show superiority of the proposed model averaging method over some commonly used variable selection methods. In addition, the proposed method is applied to a mercury data set for lakes in Maine.

Original languageEnglish
Pages (from-to)23-35
Number of pages13
JournalEnvironmental and Ecological Statistics
Volume19
Issue number1
DOIs
Publication statusPublished - 2012 Mar 1

Fingerprint

Model Averaging
Information Criterion
Penalty
Predictors
Variable Selection
Spatial Prediction
Mercury
Averaging Method
Dependent Data
Differentiable
Regression
Unstable
Numerical Experiment
Model
Estimate
Information criterion
Model averaging
lake
prediction
Variable selection

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Environmental Science(all)
  • Statistics, Probability and Uncertainty

Cite this

@article{a938efeb17c640ac88bbe53226ebb468,
title = "Geostatistical model averaging based on conditional information criteria",
abstract = "Variable selection in geostatistical regression is an important problem, but has not been well studied in the literature. In this paper, we focus on spatial prediction and consider a class of conditional information criteria indexed by a penalty parameter. Instead of applying a fixed criterion, which leads to an unstable predictor in the sense that it is discontinuous with respect to the response variables due to that a small change in the response may cause a different model to be selected, we further stabilize the predictor by local model averaging, resulting in a predictor that is not only continuous but also differentiable even after plugging-in estimated model parameters. Then Stein's unbiased risk estimate is applied to select the penalty parameter, leading to a data-dependent penalty that is adaptive to the underlying model. Some numerical experiments show superiority of the proposed model averaging method over some commonly used variable selection methods. In addition, the proposed method is applied to a mercury data set for lakes in Maine.",
author = "Chun-Shu Chen and Huang, {Hsin Cheng}",
year = "2012",
month = "3",
day = "1",
doi = "10.1007/s10651-011-0171-2",
language = "English",
volume = "19",
pages = "23--35",
journal = "Environmental and Ecological Statistics",
issn = "1352-8505",
publisher = "Springer Netherlands",
number = "1",

}

Geostatistical model averaging based on conditional information criteria. / Chen, Chun-Shu; Huang, Hsin Cheng.

In: Environmental and Ecological Statistics, Vol. 19, No. 1, 01.03.2012, p. 23-35.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Geostatistical model averaging based on conditional information criteria

AU - Chen, Chun-Shu

AU - Huang, Hsin Cheng

PY - 2012/3/1

Y1 - 2012/3/1

N2 - Variable selection in geostatistical regression is an important problem, but has not been well studied in the literature. In this paper, we focus on spatial prediction and consider a class of conditional information criteria indexed by a penalty parameter. Instead of applying a fixed criterion, which leads to an unstable predictor in the sense that it is discontinuous with respect to the response variables due to that a small change in the response may cause a different model to be selected, we further stabilize the predictor by local model averaging, resulting in a predictor that is not only continuous but also differentiable even after plugging-in estimated model parameters. Then Stein's unbiased risk estimate is applied to select the penalty parameter, leading to a data-dependent penalty that is adaptive to the underlying model. Some numerical experiments show superiority of the proposed model averaging method over some commonly used variable selection methods. In addition, the proposed method is applied to a mercury data set for lakes in Maine.

AB - Variable selection in geostatistical regression is an important problem, but has not been well studied in the literature. In this paper, we focus on spatial prediction and consider a class of conditional information criteria indexed by a penalty parameter. Instead of applying a fixed criterion, which leads to an unstable predictor in the sense that it is discontinuous with respect to the response variables due to that a small change in the response may cause a different model to be selected, we further stabilize the predictor by local model averaging, resulting in a predictor that is not only continuous but also differentiable even after plugging-in estimated model parameters. Then Stein's unbiased risk estimate is applied to select the penalty parameter, leading to a data-dependent penalty that is adaptive to the underlying model. Some numerical experiments show superiority of the proposed model averaging method over some commonly used variable selection methods. In addition, the proposed method is applied to a mercury data set for lakes in Maine.

UR - http://www.scopus.com/inward/record.url?scp=84858276254&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84858276254&partnerID=8YFLogxK

U2 - 10.1007/s10651-011-0171-2

DO - 10.1007/s10651-011-0171-2

M3 - Article

AN - SCOPUS:84858276254

VL - 19

SP - 23

EP - 35

JO - Environmental and Ecological Statistics

JF - Environmental and Ecological Statistics

SN - 1352-8505

IS - 1

ER -