Fig. 2: MOSAIKS models perform well predicting socioeconomic indicators, including at downscaled resolution. | Nature Communications

Fig. 2: MOSAIKS models perform well predicting socioeconomic indicators, including at downscaled resolution.

From: Global high-resolution estimates of the UN Human Development Index using satellite imagery and machine learning

Fig. 2

A Observed and predicted HDI at the province level. Note that the within-country variation is smaller than the across-country variation, as illustrated by France, in yellow, and Ethiopia, in pink. B The same as (A), evaluating within-country variation. Provincial deviations from the country mean for France and Ethiopia are now centered at 0, and the model is evaluated on how well it can differentiate provinces that are relatively well and worse-off within countries. C, D Observed and predicted municipal HDI data in Mexico, Brazil and Indonesia. E–G Observed and predicted International Wealth Index (IWI) at the Demographic and Health Surveys (DHS) cluster level. H–J Observed and predicted nighttime lights (NL) at the municipality level. The vertical streaking in (H–J) are caused by countries that have very spatially dense municipalities (Supplementary Information S6.3). Predictions are anchored to known country or provincial means (Methods 9). All predictions are for the year 2019.

Back to article page