The Effect of Regional Characteristics and Relationship Among Locations In Air Pollution Using Spatial Autoregressive (SAR) and Spatial Durbin Models (SDM)

Abstract

Bantul Regency in Special Region of Yogyakarta, which consists of 17 sub-districts, is an area that continues to grow and has increased population activities. In 2014, 16 sub-districts had been contaminated by air quality. To find out things that can affect air pollution in the area, this study conduct spatial model. This method is used as an alternative of OLS a method that does not meet assumptions when used in the case of spatial data. The reason for using spatial model is because there is an autocorrelation in air quality among sub-districts. By OLS regression model, a significant factor influencing is the number of villages according to the type of transportation infrastructure. However, this model does not pay attention to the geographical location factor and the residual assumption that normal distribution is not met. The results of spatial model is Spatial Durbin Model (SDM) have a good performance because have a small AIC than SAR. In SDM, the variables that significant effect on air pollution were population density, transportation infrastructure, lag of population density, lag of transportation infrastructure, and lag of air pollution. Sub district will have high of air pollutan if have high of population density and transportation infrastructure, also close to other sub districts which also have high population density and transportation infrastructure.