The Use of Geographically Weighted Regression for Spatial Prediction

An Evaluation of Models Using Simulated Data Sets

P. Harris, Stewart Fotheringham, R. Crespo, M. Charlton

Research output: Contribution to journalArticle

85 Citations (Scopus)

Abstract

Increasingly, the geographically weighted regression (GWR) model is being used for spatial prediction rather than for inference. Our study compares GWR as a predictor to (a) its global counterpart of multiple linear regression (MLR); (b) traditional geostatistical models such as ordinary kriging (OK) and universal kriging (UK), with MLR as a mean component; and (c) hybrids, where kriging models are specified with GWR as a mean component. For this purpose, we test the performance of each model on data simulated with differing levels of spatial heterogeneity (with respect to data relationships in the mean process) and spatial autocorrelation (in the residual process). Our results demonstrate that kriging (in a UK form) should be the preferred predictor, reflecting its optimal statistical properties. However the GWR-kriging hybrids perform with merit and, as such, a predictor of this form may provide a worthy alternative to UK for particular (non-stationary relationship) situations when UK models cannot be reliably calibrated. GWR predictors tend to perform more poorly than their more complex GWR-kriging counterparts, but both GWR-based models are useful in that they provide extra information on the spatial processes generating the data that are being predicted.

Original languageEnglish (US)
Pages (from-to)657-680
Number of pages24
JournalMathematical Geosciences
Volume42
Issue number6
DOIs
StatePublished - 2010
Externally publishedYes

Fingerprint

Spatial Prediction
Universal Kriging
kriging
Regression
Kriging
Predictors
Evaluation
prediction
Multiple Linear Regression
Model
Ordinary Kriging
Spatial Autocorrelation
Spatial Heterogeneity
Process Mean
Spatial Process
Statistical property
evaluation
Regression Model
Tend
autocorrelation

Keywords

  • GWR
  • Kriging
  • Relationship heterogeneity
  • Relationship nonstationarity
  • Spatial interpolation

ASJC Scopus subject areas

  • Mathematics (miscellaneous)
  • Earth and Planetary Sciences(all)

Cite this

The Use of Geographically Weighted Regression for Spatial Prediction : An Evaluation of Models Using Simulated Data Sets. / Harris, P.; Fotheringham, Stewart; Crespo, R.; Charlton, M.

In: Mathematical Geosciences, Vol. 42, No. 6, 2010, p. 657-680.

Research output: Contribution to journalArticle

@article{d2970d5363074d1aa237ee1931344316,
title = "The Use of Geographically Weighted Regression for Spatial Prediction: An Evaluation of Models Using Simulated Data Sets",
abstract = "Increasingly, the geographically weighted regression (GWR) model is being used for spatial prediction rather than for inference. Our study compares GWR as a predictor to (a) its global counterpart of multiple linear regression (MLR); (b) traditional geostatistical models such as ordinary kriging (OK) and universal kriging (UK), with MLR as a mean component; and (c) hybrids, where kriging models are specified with GWR as a mean component. For this purpose, we test the performance of each model on data simulated with differing levels of spatial heterogeneity (with respect to data relationships in the mean process) and spatial autocorrelation (in the residual process). Our results demonstrate that kriging (in a UK form) should be the preferred predictor, reflecting its optimal statistical properties. However the GWR-kriging hybrids perform with merit and, as such, a predictor of this form may provide a worthy alternative to UK for particular (non-stationary relationship) situations when UK models cannot be reliably calibrated. GWR predictors tend to perform more poorly than their more complex GWR-kriging counterparts, but both GWR-based models are useful in that they provide extra information on the spatial processes generating the data that are being predicted.",
keywords = "GWR, Kriging, Relationship heterogeneity, Relationship nonstationarity, Spatial interpolation",
author = "P. Harris and Stewart Fotheringham and R. Crespo and M. Charlton",
year = "2010",
doi = "10.1007/s11004-010-9284-7",
language = "English (US)",
volume = "42",
pages = "657--680",
journal = "Mathematical Geosciences",
issn = "1874-8961",
publisher = "Springer Netherlands",
number = "6",

}

TY - JOUR

T1 - The Use of Geographically Weighted Regression for Spatial Prediction

T2 - An Evaluation of Models Using Simulated Data Sets

AU - Harris, P.

AU - Fotheringham, Stewart

AU - Crespo, R.

AU - Charlton, M.

PY - 2010

Y1 - 2010

N2 - Increasingly, the geographically weighted regression (GWR) model is being used for spatial prediction rather than for inference. Our study compares GWR as a predictor to (a) its global counterpart of multiple linear regression (MLR); (b) traditional geostatistical models such as ordinary kriging (OK) and universal kriging (UK), with MLR as a mean component; and (c) hybrids, where kriging models are specified with GWR as a mean component. For this purpose, we test the performance of each model on data simulated with differing levels of spatial heterogeneity (with respect to data relationships in the mean process) and spatial autocorrelation (in the residual process). Our results demonstrate that kriging (in a UK form) should be the preferred predictor, reflecting its optimal statistical properties. However the GWR-kriging hybrids perform with merit and, as such, a predictor of this form may provide a worthy alternative to UK for particular (non-stationary relationship) situations when UK models cannot be reliably calibrated. GWR predictors tend to perform more poorly than their more complex GWR-kriging counterparts, but both GWR-based models are useful in that they provide extra information on the spatial processes generating the data that are being predicted.

AB - Increasingly, the geographically weighted regression (GWR) model is being used for spatial prediction rather than for inference. Our study compares GWR as a predictor to (a) its global counterpart of multiple linear regression (MLR); (b) traditional geostatistical models such as ordinary kriging (OK) and universal kriging (UK), with MLR as a mean component; and (c) hybrids, where kriging models are specified with GWR as a mean component. For this purpose, we test the performance of each model on data simulated with differing levels of spatial heterogeneity (with respect to data relationships in the mean process) and spatial autocorrelation (in the residual process). Our results demonstrate that kriging (in a UK form) should be the preferred predictor, reflecting its optimal statistical properties. However the GWR-kriging hybrids perform with merit and, as such, a predictor of this form may provide a worthy alternative to UK for particular (non-stationary relationship) situations when UK models cannot be reliably calibrated. GWR predictors tend to perform more poorly than their more complex GWR-kriging counterparts, but both GWR-based models are useful in that they provide extra information on the spatial processes generating the data that are being predicted.

KW - GWR

KW - Kriging

KW - Relationship heterogeneity

KW - Relationship nonstationarity

KW - Spatial interpolation

UR - http://www.scopus.com/inward/record.url?scp=77954860974&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77954860974&partnerID=8YFLogxK

U2 - 10.1007/s11004-010-9284-7

DO - 10.1007/s11004-010-9284-7

M3 - Article

VL - 42

SP - 657

EP - 680

JO - Mathematical Geosciences

JF - Mathematical Geosciences

SN - 1874-8961

IS - 6

ER -