A geographically weighted artificial neural network

J Hagenauer; M Helbich

doi:10.1080/13658816.2021.1871618

A geographically weighted artificial neural network

J Hagenauer, M Helbich

Urban Accessibility and Social Inclusion

Research output: Contribution to journal › Article › Academic › peer-review

Abstract

While recent developments have extended geographically weighted regression (GWR) in many directions, it is usually assumed that the relationships between the dependent and the independent variables are linear. In practice, however, it is often the case that variables are nonlinearly associated. To address this issue, we propose a geographically weighted artificial neural network (GWANN). GWANN combines geographical weighting with artificial neural networks, which are able to learn complex nonlinear relationships in a data-driven manner without assumptions. Using synthetic data with known spatial characteristics and a real-world case study, we compared GWANN with GWR. While the results for the synthetic data show that GWANN performs better than GWR when the relationships within the data are nonlinear and their spatial variance is high, the results based on the real-world data demonstrate that the performance of GWANN can also be superior in a practical setting.

Original language	English
Article number	1871618
Pages (from-to)	215-235
Number of pages	21
Journal	International Journal of Geographical Information Science
Volume	36
Issue number	2
Early online date	2021
DOIs	https://doi.org/10.1080/13658816.2021.1871618
Publication status	Published - 2022

Bibliographical note

Funding Information:
This research is an outcome of the first author?s research stay at Utrecht University in the course of the NEEDS project. The NEEDS project was funded by the European Research Council (ERC) under the European Union?s Horizon 2020 research and innovation program [Grant Agreement No. 714993]. The funders had no role in the study design, data collection and analysis, interpretation, or dissemination. We thank the anonymous reviewers for their constructive comments, which greatly improved this article. We also acknowledge UniCredit Bank Austria AG, in particular Wolfgang Brunauer, for providing the housing dataset. The opinions expressed by the authors do not reflect the official viewpoint of UniCredit Bank Austria AG.

Publisher Copyright:
© 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.

Keywords

Geographically weighted regression
artificial neural network; spatial heterogeneity; nonlinear relationships; spatial prediction

Access to Document

10.1080/13658816.2021.1871618Licence: CC BY-NC-ND

A geographically weighted artificial neural networkFinal published version, 2.03 MBLicence: CC BY-NC-ND

Cite this

@article{da94e66f2d364e0e89267d3af820c9cd,

title = "A geographically weighted artificial neural network",

abstract = "While recent developments have extended geographically weighted regression (GWR) in many directions, it is usually assumed that the relationships between the dependent and the independent variables are linear. In practice, however, it is often the case that variables are nonlinearly associated. To address this issue, we propose a geographically weighted artificial neural network (GWANN). GWANN combines geographical weighting with artificial neural networks, which are able to learn complex nonlinear relationships in a data-driven manner without assumptions. Using synthetic data with known spatial characteristics and a real-world case study, we compared GWANN with GWR. While the results for the synthetic data show that GWANN performs better than GWR when the relationships within the data are nonlinear and their spatial variance is high, the results based on the real-world data demonstrate that the performance of GWANN can also be superior in a practical setting.",

keywords = "Geographically weighted regression, artificial neural network; spatial heterogeneity; nonlinear relationships; spatial prediction",

author = "J Hagenauer and M Helbich",

note = "Funding Information: This research is an outcome of the first author?s research stay at Utrecht University in the course of the NEEDS project. The NEEDS project was funded by the European Research Council (ERC) under the European Union?s Horizon 2020 research and innovation program [Grant Agreement No. 714993]. The funders had no role in the study design, data collection and analysis, interpretation, or dissemination. We thank the anonymous reviewers for their constructive comments, which greatly improved this article. We also acknowledge UniCredit Bank Austria AG, in particular Wolfgang Brunauer, for providing the housing dataset. The opinions expressed by the authors do not reflect the official viewpoint of UniCredit Bank Austria AG. Publisher Copyright: {\textcopyright} 2021 The Author(s). Published by Informa UK Limited, trading as Taylor \& Francis Group.",

year = "2022",

doi = "10.1080/13658816.2021.1871618",

language = "English",

volume = "36",

pages = "215--235",

journal = "International Journal of Geographical Information Science",

issn = "1365-8816",

publisher = "Taylor and Francis Ltd.",

number = "2",

}

TY - JOUR

T1 - A geographically weighted artificial neural network

AU - Hagenauer, J

AU - Helbich, M

N1 - Funding Information: This research is an outcome of the first author?s research stay at Utrecht University in the course of the NEEDS project. The NEEDS project was funded by the European Research Council (ERC) under the European Union?s Horizon 2020 research and innovation program [Grant Agreement No. 714993]. The funders had no role in the study design, data collection and analysis, interpretation, or dissemination. We thank the anonymous reviewers for their constructive comments, which greatly improved this article. We also acknowledge UniCredit Bank Austria AG, in particular Wolfgang Brunauer, for providing the housing dataset. The opinions expressed by the authors do not reflect the official viewpoint of UniCredit Bank Austria AG. Publisher Copyright: © 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.

PY - 2022

Y1 - 2022

N2 - While recent developments have extended geographically weighted regression (GWR) in many directions, it is usually assumed that the relationships between the dependent and the independent variables are linear. In practice, however, it is often the case that variables are nonlinearly associated. To address this issue, we propose a geographically weighted artificial neural network (GWANN). GWANN combines geographical weighting with artificial neural networks, which are able to learn complex nonlinear relationships in a data-driven manner without assumptions. Using synthetic data with known spatial characteristics and a real-world case study, we compared GWANN with GWR. While the results for the synthetic data show that GWANN performs better than GWR when the relationships within the data are nonlinear and their spatial variance is high, the results based on the real-world data demonstrate that the performance of GWANN can also be superior in a practical setting.

AB - While recent developments have extended geographically weighted regression (GWR) in many directions, it is usually assumed that the relationships between the dependent and the independent variables are linear. In practice, however, it is often the case that variables are nonlinearly associated. To address this issue, we propose a geographically weighted artificial neural network (GWANN). GWANN combines geographical weighting with artificial neural networks, which are able to learn complex nonlinear relationships in a data-driven manner without assumptions. Using synthetic data with known spatial characteristics and a real-world case study, we compared GWANN with GWR. While the results for the synthetic data show that GWANN performs better than GWR when the relationships within the data are nonlinear and their spatial variance is high, the results based on the real-world data demonstrate that the performance of GWANN can also be superior in a practical setting.

KW - Geographically weighted regression

KW - artificial neural network; spatial heterogeneity; nonlinear relationships; spatial prediction

UR - http://www.scopus.com/inward/record.url?scp=85100716480&partnerID=8YFLogxK

U2 - 10.1080/13658816.2021.1871618

DO - 10.1080/13658816.2021.1871618

M3 - Article

SN - 1365-8816

VL - 36

SP - 215

EP - 235

JO - International Journal of Geographical Information Science

JF - International Journal of Geographical Information Science

IS - 2

M1 - 1871618

ER -

A geographically weighted artificial neural network

Abstract

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this