The Global Poverty Map

The Data-Intensive Development Lab (DIDL) has developed a method of estimating micro-regional levels of poverty based on satellite imagery, and mobile phone and social network usage. The DIDL has then used this machine-learning derived data to create an interactive poverty map visualizing micro-regional poverty levels in Low and Middle-Income Countries (LMICs). The maps also use poverty and wealth data gathered from standardized and publicly-available Demographic and Health Surveys carried out in these LMICs.

To arrive at the micro-regional estimates of poverty levels the DIDL starts with the data derived from Demographic and Health Surveys. Then they analyze satellite imagery of the region, looking for indicators of local living conditions, such as the size of farm plots, the quality of roads and the quality of roofing materials. Night-time satellite imagery is also used to derive data on electricity use. Telecommunications infrastructure is derived from analyzing the density of mobile cellular towers, the number of WiFi access points and from the levels of customer use of mobile phone networks.

By combining the data from surveys with the machine learning derived data from satellite imagery and from telecommunications infrastructure DIDL are able to estimate wealth levels in every 2.4KM micro-region on Earth.

You can explore this data for yourself on the interactive Global Poverty Map, which shows these estimates of wealth and poverty in micro-regions around the world. The map visualizes population density using height and poverty levels are shown using color. The taller micro-regions have the highest population density and the brown colored squares have relatively higher wealth.