GRID3
Machine learning fills in missing population maps, enabling governments to plan vaccinations, censuses, and other public services more precisely.
About
GRID3 works with African governments on maps of population, settlements, borders, and amenities. For areas with outdated or incomplete censuses, its models combine sample counts with satellite imagery, buildings, and other spatial data to produce population estimates in grid cells of roughly 100 by 100 meters, including uncertainty margins. In Nigeria, such maps were used for microplanning vaccination campaigns and contributed to the efforts after which the African region was certified free of wild poliovirus in 2020. A model map remains an estimate and cannot replace an actual census and local verification.
AI factor
A human cannot reliably infer where all residents live from a few counts and millions of map cells. Machine-learning and Bayesian models learn the relationship between counted population, built structures, and surroundings, then estimate uncounted areas; local statistical services verify and use the results.
Sources
Inspiration sourced from AI For Good. Not an Arnaldus customer case.