Estimating Geographically Weighted Negative Binominal Regression (GWNBR) Model and Mapping Spatial Patterns on Anemia Data in Indonesia

Abstract: The Geographically Weighted Negative Binomial Regression (GWNBR) model incorporates spatial weighting based on the latitude and longitude of observation locations and is particularly effective in addressing issues related to overdispersion. In this study, the model was applied to actual data concerning anemia cases among women of childbearing age (WCA) across 33 provinces in Indonesia. By employing the GWNBR approach, researchers were able to estimate parameters and analyze the spatial variation in regression coefficients. The analysis revealed that in each province, at least one predictor variable had a statistically significant influence on the response variable. In the central and eastern regions of Indonesia, an increase in anemia cases per 100 WCA was positively associated with the number of WCA affected by pneumonia, malaria, and hepatitis. Meanwhile, in western Indonesia, the rise in anemia prevalence among 100 WCA was positively correlated with the number of WCA residing in rural areas (notably in Java and Kalimantan), as well as those affected by acute respiratory infections (in Sumatra and Kalimantan) and tuberculosis.