Localizing the SDG Index with Machine Learning and Satellite Imagery

Progress towards the SDGs has been slow, and significant data gaps hinder the ability to track SDG performance, particularly at the local level. The SDG Transformation Center addresses these gaps by generating new geospatial datasets and refining existing indicators to better capture regional and demographic disparities. Using artificial neural networks and high-resolution satellite imagery, we built on existing methodologies to downscale SDG Index scores from national to subnational levels. This model helps to identify aggregation biases in national SDG indicators, revealing that nearly 43% of the global population may be overlooked when only national averages are considered. Country-specific subnational analyses highlight disparities in SDG progress, especially between urban centers and rural regions. While these findings provide valuable insights, future efforts should focus on refining the model with locally validated SDG data to further improve accuracy and support evidence-based, geographically targeted policies for sustainable development.