Geospatial micro-estimates of slum populations in 129 Global South countries using machine learning and public data

Reliable estimation of populations living in slums or slum-like conditions is crucial for urban planning, humanitarian resource allocation, and human well-being improvement. We generate the micro-estimate of slum population at a neighborhood level (~3.63 arc-minutes, preserving the privacy of vulnerable people) for 129 Global South countries in 2018. The estimates are built based on the Sustainable Development Goals 11.1 indicator framework and machine learning algorithms to heterogeneous data from household-based surveys and satellite images, as well as grided population data. Our models explain 82–95% of the variation in within-country spatial prediction and 45–91% in country-level holdout validation, corresponding to median R² values of 0.89 and 0.85, respectively. These results indicate strong predictive performance and remain broadly comparable with, and in some cases higher than, previously reported benchmarks. Cross-comparison with independent data sources at multi-scales suggest that our approach can yield reliable and consistent slum population estimates.