Closing the Data Divide: Machine Learning Approaches for Understanding Livelihoods of the Poor Using Unconventional Data Sources
Principal Investigator
Co-Investigators
Abstract
New developments in sensing technologies are creating many cheap, unconventional – but also unstructured and noisy - data streams that should contain information relevant to poverty and hunger. New computational methods are therefore needed for transforming large amounts of unstructured data into actionable insights. The goal of this research project is to deliver a new generation of poverty measures, based on a combination of ubiquitous data streams and novel machine learning approaches. We will first develop and test new approaches to measure poverty and hunger, and than use these new measures to study the determinants of the outcomes. We anticipate that the results will be truly transformational for a variety of scientific disciplines and questions, and particularly affect our understanding of the economics of developing countries.