Exploiting mobile phone data for multi-category land use classification in Africa

Abstract

In the context of Smart Africa Initiative, we present a method to infer multiple land use in Africa. Such information is usually scarce in developing countries due to the constrained resources. Timely land use information is a critical input to smart urban planning that improves efficiency for the public to access to resources. The mobile phone usage is almost universal, which creates a valuable data source for land use inference. In this paper, we demonstrate that the temporal mobile phone call pattern and call network features can be combined to infer ten-category land use including residential, commercial-industrial/office, commercial-business/retail/leisure, high- and low- density commercial, high- and low- density residential, mixed land use areas as well as commercial and residential hubs of the city. In low income countries where land use surveys are rare, our approach create an alternative for measuring land use.