The Step Construction of Geographically Weighted Panel Regression in Air Polluter Standard Index (APSI) Data

Budi Warsito, Hasbi Yasin, Dwi Ispriyanti, Arief Rachman Hakim, Hadiyanto, Maryono, Budi Warsito
2018 E3S Web of Conferences  
Geographically Weighted Panel Regression or GWPR is a local linear regression model that combines GWR model and panel data regression model with considering spatial effect, especially spatial heterogeneity problem. This article is focused on the soft computation of GWPR model using Fixed Effect Model (FEM). Parameter estimation in GWPR is obtain by Weighted Least Squares (WLS) methods and the resulting model for each location will be different from one to another. This study will compare the
more » ... edeffect GWPR model with several weighting functions. The best model is determined based on the biggest coefficient of determination (R 2 ) value. In this study, the model is applied in the Air Polluter Standard Index (APSI) in Surabaya City, East Java. The results of this study indicate that Fixed Effect GWPR model with a fixed exponential kernel weighting function is the best model to describe the APSI because it has the smallest AIC.
doi:10.1051/e3sconf/20187312006 fatcat:khpkxvp4anf2dorstdjkspohf4