Non-path dependent urban growth potential mapping using a data-driven evidential belief function

Bibliographic Data

ID21248087
AuthorsReza Arasteh (University of Tehran), Rahim Ali Abbaspour (0000-0002-7133-3844, University of Tehran, corresponding author), Abdolrassoul Salmanmahiny (0000-0002-5188-7356, Gorgan University of Agricultural Sciences and Natural Resources)
Year2021
Volume48
Issue3
Pages555-573
Publication date2021-03-01
Peer ReviewedYes
Open AccessYes
TypeARTICLE
VenueEnvironment and Planning B Urban Analytics and City Science (JOURNAL)
Journal identifiersISSN: 2399-8083 • E-ISSN: 2399-8091
PublisherSAGE Publications (PUBLISHER • US)
DOI10.1177/2399808319880219
OpenAlexW2980250899
LanguageEN
References cited34

Improper urbanization and its environmental impacts have imposed many problems to humanity. Recently, numerous studies have been conducted using different methods to understand and manage spatial and temporal changes in urbanization. In this study, the capability of the data-driven evidential belief function model as a non-path dependent urban growth potential mapping method was evaluated. Using this approach, the conventional trend-based urban growth prediction procedure is transformed into a data integration task through which the potential locations for urban sprawl in response to multiple environmental and anthropogenic variables could be determined. Therefore, true knowledge about urban growth conditioning factors and their quantitative relationships with built-up areas can be obtained. The multivariate-based logistic regression model as a well-known urban growth modelling method was employed to check the efficiency and validity of the proposed evidential belief function model. Furthermore, a hybrid approach based on the logistic regression model results coupled with the data-driven evidential belief function model was developed. The validation results using the relative operating characteristic method indicated that the evidential belief function, logistic regression, and the hybrid methods’ accuracy were 91.81, 84.72, and 92.34%, respectively. Therefore, it can be concluded that while the proposed evidential belief function and hybrid methods for non-path dependent urban growth potential mapping yielded approximately equal results, both of them outperformed the logistic regression model which is an indication of their reliability and accuracy. The proposed evidential belief function and hybrid methods are best suited for integration in different environmental and socio-economic scenarios enhancing the models for urban allocation tasks. In this way, local communities and policy-makers can make smarter decisions

Data mining · Decision support system · Econometrics · Evidential reasoning approach · Logistic function · Logistic regression · Machine learning · Regression analysis · Statistics · Urban planning · Urban sprawl · Computer Science · Engineering · Land Use and Ecosystem Services · Mathematics · Remote Sensing and Land Use · Soil and Land Suitability Analysis

Tools

Open DOISci-Hub