Exploring image-based indicators of crime and economic well-being in sub-saharan Africa

Abstract

Crime and economic well-being are important factors to consider when assessing the stability of a region. Gaining insight into these factors can be expensive and perilous, depending on the accessibility of the country and the cooperation of the government. However, the abundance of satellite imagery provides an avenue to analyze crime and economic indicators. Using social science theory, we examine the ability of physical structures and surroundings to indicate societal issues. Draper's previous studies explored the ability of satellite image features to access indicators of governance and economic well-being in Afghanistan and Sub-Saharan Africa. Such research has provided promising insight into identifying feature sets from satellite imagery that are useful in predicting the economic and political landscape of the country. Past work also described an automated technique to quantify the amount of built-up areas in a village. In this research, we extend the quantification to multiple types of built-up areas. This quantification adds another layer of description to our library of spatial characteristics. We then use these new feature sets to deepen and enhance the understanding of indicators of crime and economic-well-being. We use survey data to access the improvement of the predictive power using the additional feature sets.