<![CDATA[Sustainable development agenda requires accurate and granular data support. However, accurate and granular data are sparse, due to the limitation of various resources. This study explores the possibility of producing Gross Domestic Product (GDP) estimates at the more granular level, i.e. at 2.4-km grid cell (microregional GDP). To achieve the purpose, this study utilizes several high-resolution geospatial predictors, such as night-time lights, land cover, topography and the location of economic activities. This study covers Java islands only. The estimates of microregional GDP are produced using several machine learning models, such as LASSO, Elastic Net, Support Vector Machine and Random Forest. The results showed that Random Forest was the best model for estimating the microregional GDP, where the night-time lights was the best predictor. This study also validated the results using independent data sources, such as the Relative Wealth Index. Validation results showed that the microregional GDP estimates were quite reliable.]]>
Copyrights © 2022
