A statistical analysis for geographical weighted regression

9th IGRSM International Conference and Exhibition on Geospatial & Remote Sensing

地球科学;计算机科学

Nur Edayu, Z.^1 ; Syerrina, Z.^1

School of Informatics and Applied Mathematics, University Malaysia Terengganu, Kuala Terengganu

21300, Malaysia^1

关键词: Geographical weighted regressions;    Geographically weighted regression models;    Hypothesis testing;    Ordinary least squares;    R programming;    Spatial non-stationarity;    Statistical inference;    Validation process;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/169/1/012105/pdf
DOI  :  10.1088/1755-1315/169/1/012105

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

Geographically Weighted Regression (GWR) model provides a spatial nonstationarity analysis for regression that gives different relationships to exist at different points in space. Moreover, GWR model provides a better specification to model rather than Ordinary Least Square Regression (OLS). It can be easily applied to spatial data by using the R packages that are available in R programming or GWR4 software. Though, the statistical test studies for GWR infrequently explored due to its complexity problem. Currently, the GWR studies are only limited for exploration on the relationship between the variables in study field. While the validation process for hypothesis testing of GWR model is difficult to conduct due to multiple parameter estimations for each location. The main purpose of this paper is to summaries a statistical inferences analysis for GWR model. The hypothesis testing will be derived by testing the variation of the parameter in the model. This study gives a better understanding of the different type of statistical inferences procedures that can be applied by researchers and this encourage more application for statistical inferences study from different fields.