Detecting Social Networks with Satellite Imagery

Development interventions can improve effectiveness by targeting individuals more central in a network. However, mapping social networks within rural villages comes with a high cost. This project offers a less costly alternative by taking advantage of preexisting satellite data and spatial modeling techniques. Specifically, it links detailed network data from over 200 villages in Malawi and India with satellite data. Doing so explores the possibility of predicting a village’s social network structure using only satellite data, a prediction algorithm that could potentially be automated and require no on-the-ground survey. The project also aims to discover whether the prediction quality generated can increase the efficacy of rural development interventions. Results forthcoming.