Spatial contextual classification and prediction models for mining geospatial data

Shashi Shekhar, Paul R. Schrater, Ranga R. Vatsavai, Weili Wu, Sanjay Chawla

Research output: Contribution to journalArticle

98 Citations (Scopus)

Abstract

Modeling spatial context (e.g., autocorrelation) is a key challenge in classification problems that arise in geospatial domains. Markov random fields (MRF) is a popular model for incorporating spatial context into image segmentation and land-use classification problems. The spatial autoregression (SAR) model, which is an extension of the classical regression model for incorporating spatial dependence, is popular for prediction and classification of spatial data in regional economics, natural resources, and ecological studies. There is little literature comparing these alternative approaches to facilitate the exchange of ideas (e.g., solution procedures). We argue that the SAR model makes more restrictive assumptions about the distribution of feature values and class boundaries that MRF. The relationship between SAR and MRF is analogous to the relationship between regression and Bayesian classifiers. This paper provides comparisons between the two models using a probabilistic and an experimental framework.

Original languageEnglish
Pages (from-to)174-188
Number of pages15
JournalIEEE Transactions on Multimedia
Volume4
Issue number2
DOIs
Publication statusPublished - Jun 2002
Externally publishedYes

Fingerprint

Data mining
Natural resources
Image segmentation
Autocorrelation
Land use
Ion exchange
Classifiers
Economics

Keywords

  • Markov random fields (MRF)
  • Spatial autoregression (SAR)
  • Spatial context
  • Spatial data mining

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Information Systems
  • Computer Graphics and Computer-Aided Design
  • Software

Cite this

Spatial contextual classification and prediction models for mining geospatial data. / Shekhar, Shashi; Schrater, Paul R.; Vatsavai, Ranga R.; Wu, Weili; Chawla, Sanjay.

In: IEEE Transactions on Multimedia, Vol. 4, No. 2, 06.2002, p. 174-188.

Research output: Contribution to journalArticle

Shekhar, Shashi ; Schrater, Paul R. ; Vatsavai, Ranga R. ; Wu, Weili ; Chawla, Sanjay. / Spatial contextual classification and prediction models for mining geospatial data. In: IEEE Transactions on Multimedia. 2002 ; Vol. 4, No. 2. pp. 174-188.
@article{f366224d6c304d78921120833cca66e4,
title = "Spatial contextual classification and prediction models for mining geospatial data",
abstract = "Modeling spatial context (e.g., autocorrelation) is a key challenge in classification problems that arise in geospatial domains. Markov random fields (MRF) is a popular model for incorporating spatial context into image segmentation and land-use classification problems. The spatial autoregression (SAR) model, which is an extension of the classical regression model for incorporating spatial dependence, is popular for prediction and classification of spatial data in regional economics, natural resources, and ecological studies. There is little literature comparing these alternative approaches to facilitate the exchange of ideas (e.g., solution procedures). We argue that the SAR model makes more restrictive assumptions about the distribution of feature values and class boundaries that MRF. The relationship between SAR and MRF is analogous to the relationship between regression and Bayesian classifiers. This paper provides comparisons between the two models using a probabilistic and an experimental framework.",
keywords = "Markov random fields (MRF), Spatial autoregression (SAR), Spatial context, Spatial data mining",
author = "Shashi Shekhar and Schrater, {Paul R.} and Vatsavai, {Ranga R.} and Weili Wu and Sanjay Chawla",
year = "2002",
month = "6",
doi = "10.1109/TMM.2002.1017732",
language = "English",
volume = "4",
pages = "174--188",
journal = "IEEE Transactions on Multimedia",
issn = "1520-9210",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "2",

}

TY - JOUR

T1 - Spatial contextual classification and prediction models for mining geospatial data

AU - Shekhar, Shashi

AU - Schrater, Paul R.

AU - Vatsavai, Ranga R.

AU - Wu, Weili

AU - Chawla, Sanjay

PY - 2002/6

Y1 - 2002/6

N2 - Modeling spatial context (e.g., autocorrelation) is a key challenge in classification problems that arise in geospatial domains. Markov random fields (MRF) is a popular model for incorporating spatial context into image segmentation and land-use classification problems. The spatial autoregression (SAR) model, which is an extension of the classical regression model for incorporating spatial dependence, is popular for prediction and classification of spatial data in regional economics, natural resources, and ecological studies. There is little literature comparing these alternative approaches to facilitate the exchange of ideas (e.g., solution procedures). We argue that the SAR model makes more restrictive assumptions about the distribution of feature values and class boundaries that MRF. The relationship between SAR and MRF is analogous to the relationship between regression and Bayesian classifiers. This paper provides comparisons between the two models using a probabilistic and an experimental framework.

AB - Modeling spatial context (e.g., autocorrelation) is a key challenge in classification problems that arise in geospatial domains. Markov random fields (MRF) is a popular model for incorporating spatial context into image segmentation and land-use classification problems. The spatial autoregression (SAR) model, which is an extension of the classical regression model for incorporating spatial dependence, is popular for prediction and classification of spatial data in regional economics, natural resources, and ecological studies. There is little literature comparing these alternative approaches to facilitate the exchange of ideas (e.g., solution procedures). We argue that the SAR model makes more restrictive assumptions about the distribution of feature values and class boundaries that MRF. The relationship between SAR and MRF is analogous to the relationship between regression and Bayesian classifiers. This paper provides comparisons between the two models using a probabilistic and an experimental framework.

KW - Markov random fields (MRF)

KW - Spatial autoregression (SAR)

KW - Spatial context

KW - Spatial data mining

UR - http://www.scopus.com/inward/record.url?scp=0036613147&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0036613147&partnerID=8YFLogxK

U2 - 10.1109/TMM.2002.1017732

DO - 10.1109/TMM.2002.1017732

M3 - Article

VL - 4

SP - 174

EP - 188

JO - IEEE Transactions on Multimedia

JF - IEEE Transactions on Multimedia

SN - 1520-9210

IS - 2

ER -