Geographically Weighted Quantile Regression for Assessing the Spatial Variations of the High-Risk Diarrhea on Children Under Five in Bandung, Indonesia

Abstract

Abstract. We investigate the impact of the percentage of clean water access, the percentage of handwashing habit, and the toilet category factors on the upper quantile of toddlers’ diarrhea risks in Bandung city, Indonesia using Geographically Weighted Quantile Regression model. We analyzed districts in Bandung, applying Geographically Weighted Quantile Regression with the predictors are the percentage of clean water access, hand washing habit, and toilet category. We focus on the 75th percentile ( . Breusch-Pagan test was used to detect the spatial heterogeneity. The optimum bandwidth was selected using cross-validation. The results shows that the significance, strength, and direction of the relationship between diarrhea and its risk factors are found to depend on the location. At the upper quantile τ = 0.75, the district named Panyileukan district is predicted to have the highest diarrhea risk. In this district, all of the three predictors are significantly affecting the toddlers’ diarrhea risk with the variable of percentage of houses practicing hand washing habit is observed to reduce diarrhea risk the most. In conclusion, clean water access, handwashing habits, and toilet category are the potential risk factors of high-risk childhood diarrhea. The significance, strength, and direction of the effect varies between districts. This method is powerful as it would allow the decision maker to handle the diarrhea problem aptly based