Predictors of SDG Achievement in Indonesian Local Governments: An Application of Machine Learning Techniques

This study investigates the predictors of SDGs achievement (SDGs 1 No Poverty) of kabupaten/kota (or district and city local governments) in Indonesia. The predictors were local government size, mayor profile (age, gender, and education), and citizen quality (human development index/HDI). The study utilized Slovin's formula to determine the sample size and randomly selected 230 local governments using an online randomization tool. Multiple regression analysis was conducted to determine the influence of government size, HDI, age, gender, and education of the mayors on SDG indicators related to poverty (SDG 1). To comprehend the results, the decision tree technique was applied to identify the most important factors influencing poverty in the local government samples. This study found that the predictors of SDG 1 achievement were local government size, mayor age, and citizen quality. Multiple regression shows that government size and mayor's age have a significant and positive impact, while HDI has a significant and negative impact on poverty rates. However, the decision tree analysis only documented citizen quality as a predictor of poverty rates in Indonesian local governments. The reason could be that decision tree emphasizes the strongest predictor (citizen quality) rather the combined linear effects of all variables.