Mike Prastuti, Erma Oktania Permatasari : Seasonal GSTAR-SUR Model Used Spatial Weight Normalization Statistical Inference of Partial Cross- Correlation

Mike Prastuti S.Si., M.Si.
Erma Oktania Permatasari S.Si, M.Si



Published in

The 6th Annual Basic Science International Conference

External link


Seminar Internasional


GSTAR, OLS, GLS, space-time, DBD


Generalized Space Time Autoregressive (GSTAR) is one of space-time models which allowing the autoregressive parameters to different per locations. In generally, method that usually applied to to estimate the parameters of GSTAR model is Ordinary Least Square (OLS). Parameter estimation by using OLS for GSTAR model with correlated residuals between equations will produce inefficient estimators. The method that appropriate to estimate the parameter model with correlated residuals between equations is Generalized Least Square (GLS), which is usually used in SUR model. The purpose this research is to compare between GSTAR-OLS and GSTAR-SUR model with spatial weight normalization statistical inference of partial cross-Correlation for the number of Dengue Fever (DBD) patients at three location in east java is Surabaya, Gresik, and Sidoarjo. Based the best model from two method which is used, obtained result that GSTAR-SUR model yields parameters which more efficient than GSTAR-OLS. It is showed by the smaller standard error of GSTAR-SUR estimators. Besides that, GSTAR-SUR model yields RMSE is smaller than GSTAR-OLS at locations in Gresik and Sidoarjo. While at location in Surabaya GSTAR-OLS yields RMSE is smaller than GSTAR-SUR models.