Ubiquitous Sensing for Mapping Poverty in Developing Countries
11 February 2013
Social surveys and censuses are good indication of poverty and inequality in a country. However, due to the expenses associated to data collection, the granularity and frequency of these information is often highly impacted. The usefulness of this information indeed highly decays in the developing countries where the census is collected very infrequent, failing to reflect the fast pace of changes associated with growing economy. In this paper, we propose ubiquitous sensing as a proxy for estimating socio-economic indicators and analyse aggregated volume of mobile phone calls and SMS exchanges in Ivory Coast. We discover a number of features, such as diversity, which show a strong correlation with poverty indicators. We then show how these features can be used to provide poverty estimates at a resolution much finer than previously available.