View Item 

An optimal parameters-based geographical detector model enhances geographic characteristics of explanatory variables for spatial heterogeneity analysis: cases with different types of spatial data

Access Status
In process
Authors
Song, Yongze
Wang, J.
Ge, Y.
Xu, C.
Date
2020
Type
Journal Article

Metadata
Show full item record
Citation
Song, Y. and Wang, J. and Ge, Y. and Xu, C. 2020. An optimal parameters-based geographical detector model enhances geographic characteristics of explanatory variables for spatial heterogeneity analysis: cases with different types of spatial data. GIScience and Remote Sensing.
Source Title
GIScience and Remote Sensing
DOI
ISSN
Faculty
Faculty of Humanities
School
School of Design and the Built Environment
URI
Collection
Abstract

© 2020, © 2020 Informa UK Limited, trading as Taylor & Francis Group. Spatial heterogeneity represents a general characteristic of the inequitable distributions of spatial issues. The spatial stratified heterogeneity analysis investigates the heterogeneity among various strata of explanatory variables by comparing the spatial variance within strata and that between strata. The geographical detector model is a widely used technique for spatial stratified heterogeneity analysis. In the model, the spatial data discretization and spatial scale effects are fundamental issues, but they are generally determined by experience and lack accurate quantitative assessment in previous studies. To address this issue, an optimal parameters-based geographical detector (OPGD) model is developed for more accurate spatial analysis. The optimal parameters are explored as the best combination of spatial data discretization method, break number of spatial strata, and spatial scale parameter. In the study, the OPGD model is applied in three example cases with different types of spatial data, including spatial raster data, spatial point or areal statistical data, and spatial line segment data, and an R “GD” package is developed for computation. Results show that the parameter optimization process can further extract geographical characteristics and information contained in spatial explanatory variables in the geographical detector model. The improved model can be flexibly applied in both global and regional spatial analysis for various types of spatial data. Thus, the OPGD model can improve the overall capacity of spatial stratified heterogeneity analysis. The OPGD model and its diverse solutions can contribute to more accurate, flexible, and efficient spatial heterogeneity analysis, such as spatial patterns investigation and spatial factor explorations.