SGWR: similarity and geographically weighted regression

M. Naser Lessani,Zhenlong Li,M. Naser LessaniZhenlong LiGeoinformation and Big Data Research Laboratory,Department of Geography,The Pennsylvania State University,University Park,PA,USAM. Naser Lessani is currently a PhD student in the Department of Geography at The Pennsylvania State University. His primary research interests are geospatial big data analytics,human mobility,parallel spatial computing,Machine Learning,and GIS. He contributed to conceptualization of the research idea,data analysis,code development,and writing.Zhenlong Li is an Associate Professor in the Department of Geography and Director of the Geoinformation and Big Data Research Lab at The Pennsylvania State University. His primary research field is GIScience with a focus on geospatial big data analytics,spatial computing,and geospatial AI with applications to disaster management,human mobility,and public health. He contributed to conceptualization of the research idea,data analysis,and writing.

DOI: https://doi.org/10.1080/13658816.2024.2342319

2024-04-17

International Journal of Geographical Information Science

Abstract:Geographically weighted regression (GWR) offers a local approach to modeling spatial data, considering geographical location and spatial relationships between observations. A salient feature of GWR is the emphasis on geographical proximity, in accordance with Tobler's First Law of Geography, which assumes that closer entities have a greater influence on the target location. Traditional GWR models have been augmented to consider various forms of physical distances aimed at enhancing model performance, and they often disregarded the potential influence of other data attributes, a shortcoming that extends to most GWR extensions. In this study, we introduce a novel weight matrix construction, which integrates data attribute similarity alongside the conventional geographically weighted matrix. The two weights are integrated in a manner that results in improved model performance. The proposed model, called Similarity and Geographically Weighted Regression or SGWR, was applied to five distinct datasets: housing prices, crime rates, and three health outcomes including mental health, depression, and HIV. Results show that SGWR significantly improved model performance based on several statistical measures, outperforming the global regression model and the traditional GWR.

geography, physical,computer science, information systems,information science & library science

What problem does this paper attempt to address?

This paper introduces a new method called "Similarity and Geographically Weighted Regression" (SGWR) aimed at improving the traditional Geographically Weighted Regression (GWR) model. GWR is a local regression method that models spatial data by considering the spatial relationship between geography and observations. However, traditional GWR mainly focuses on the proximity of geographic locations and ignores the impact of data attribute similarity.

The paper proposes a new way to construct the weight matrix by combining traditional geographic weight matrix and attribute-based similarity weight. Through this approach, the new model shows better performance on multiple statistical indicators, outperforming global regression models and traditional GWR. The researchers applied SGWR to analyze five different datasets, including housing prices, crime rates, and three health outcomes (mental health, depression, and HIV), and the results show that SGWR significantly improves model performance.

The paper also discusses the limitations of GWR, such as relying solely on physical distance as a measure of spatial proximity while ignoring other data attributes. SGWR, by integrating attribute similarity, is able to more accurately capture spatial patterns and consider the spatial heterogeneity of relationships between different variables. Through empirical studies in different domains, the paper demonstrates the advantages of SGWR in understanding and predicting complex spatial phenomena.