The performance of goodness of fit test procedure on geographically weighted polynomial regression model

Tolia Saifixdin, Fatmawati, Nur Chamidah

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

A development of geographically weighted regression (GWR) model has been build, namely geographically weighted polynomial regression (GWPolR) model. Because it has more parameters, it has goodness of fit measures better than the GWR model does. However, it is urgent to test statistically whether the GWPolR model is significantly better than the GWR model in describing a given data set. There has not been a research to solve this problem. The purpose of this research is to discover a goodness of fit test procedure and its performance]. Based on the residual sum of squares (RSS) of GWR and GWPolR models and the distribution theory of quadratic forms, a statistical test approach was derived here. Furthermore, performance of the goodness of fit test procedure was investigated using some simulation studies. The results showed that the test procedure works well and can select an appropriate model for a given data set.

Original languageEnglish
Title of host publicationInternational Conference on Mathematics, Computational Sciences and Statistics 2020
EditorsCicik Alfiniyah, Fatmawati, Windarto
PublisherAmerican Institute of Physics Inc.
ISBN (Electronic)9780735440739
DOIs
Publication statusPublished - 26 Feb 2021
EventInternational Conference on Mathematics, Computational Sciences and Statistics 2020, ICoMCoS 2020 - Surabaya, Indonesia
Duration: 29 Sept 2020 → …

Publication series

NameAIP Conference Proceedings
Volume2329
ISSN (Print)0094-243X
ISSN (Electronic)1551-7616

Conference

ConferenceInternational Conference on Mathematics, Computational Sciences and Statistics 2020, ICoMCoS 2020
Country/TerritoryIndonesia
CitySurabaya
Period29/09/20 → …

Fingerprint

Dive into the research topics of 'The performance of goodness of fit test procedure on geographically weighted polynomial regression model'. Together they form a unique fingerprint.

Cite this