# Residential buildings

Predict whether a building is residential or not from its footprint

# Rationale

A number of projects and companies are producing building footprint data, providing a rich source of information on human footprint. Typically, these data are unlabelled, meaning we know there is a building there, but it isn't known what type of building it is. This limits the value these data can provide, particularly for health and humanitarian programs who are often interested in understanding where people live.

# Our approach

Building on previous work, we use machine learning algorithms, trained using ground-truthed data available in OpenStreetMap, to classify buildings as residential or non-residential with high (>90%) accuracy. This allows us to provide much higher resolution information on population and numbers of buildings to visit in a given area.

# Where is this being applied

In collaboration with multiple partners, we have conducted building classification as part of malaria control efforts in southern Africa.

Think this sounds useful?

You can reach us at hello@locational.io to ask any questions, request additions or changes, or ask for a demo. We are actively developing these algorithms and would like to hear from you.