Pemodelan angka kematian bayi di Indonesia menggunakan Geographically Weighted Regression (GWR) dan Mixed Geographically Weighted Regression (MGWR)

Abstract

The Infant Mortality Rate (IMR) is fundamental indicator that reflects the health status in the surrounding community. The Infant Mortality Rate is still categorized as high in Indonesia. Therefore, this study aims to determine the appropriate model in estimating the Infant Mortality Rate (IMR) and to find out the factors that influence the IMR in Indonesia. The data in this study was secondary which obtained from the Indonesia Health Profile. The estimation was carried out using Geograpically Weigthed Regression (GWR) and Mixed Geographically Weigthed Regression (MGWR) models. The GWR model is development of regression that consider spatial factors. While the MGWR model is a combination of regression and GWR with several variables influence locally. but the rest goes globally. The result showed that the MGWR model was the best model compared to the GWR model with the lowest AIC value selection standart. The MGWR model with weighted Adactive Kernel Gaussian found that locally influencing factors were infants who were exclusively breastfed (ASI) and infants who received early initiation of breastfeeding (IMD), while globally influencing factors were infants who were given vitamin A, low birth weight (LBW) delivery services at health facilities and pregnant women receiving bloodsupplementing tables (TTD). Keywords: Adaptive of kernel Gaussian, AIC, the infant mortality rate, GWR, MGWR MSC2020: 62M10