World Data Lab specializes in creating economic and demographic models that help organizations make data-driven decisions. Our global team of data scientists, economists, demographers, and developers are passionate about combining new data sources with groundbreaking methods to push the frontiers of what is knowable.
Our Spatial Demography team creates highly granular population density estimates at the 100 square meter grid level. Such data are used by public agencies and development organizations for policy planning, project design, and census roll-out support. Moreover, private companies benefit from these models by being able to anticipate likely changes in demand in areas that are undergoing considerable change (either rapid population growth or decline).
This population mapping model produces population density information for a given geographic location at the 100 sqm grid level. The methodology involves using a dasymetric redistribution approach based on a Random Forest nonparametric model. Data sources include a combination of official statistics, geospatial data, and other ancillary data. Random Forest leads to superior results compared to other methods (such as GRUMP and AfriPop methods).
In this approach, a Random Forest framework (i.e. a “random forest” of decision trees, where each tree only uses a random subset of covariates and input data) is used to disaggregate the population data based on census data. Results from the model are then based on averages of the estimates of each tree. In order to validate results, the estimated population for the grids are summed up for the finest level in the census data, and the two numbers are then compared.
We are interested in working with clients who want to know more about population dynamics at a highly granular level. How can we put our expertise in population modeling to work for you?