The human population continues to grow, and more of us are attaining a reasonable standard of living. Unfortunately, this is not true for all.
In the U.S. alone, about 40 million people live in poverty. Supporting, in the long run, the increasing human population will be a global challenge. Unfortunately, the sectors of the population and the areas of the world that are most at risk for poverty are hard to reach, and the absence of reliable measurements has made it difficult to build models and design interventions. Yet there are many new promising sources of data that should help. These include both public data streams (e.g., social media apps, Wikipedia and Wikidata and moderate-resolution satellite imagery), as well as private-sector data (e.g., cell phone records, Facebook activity, internet search queries, drone imagery and fine-resolution satellite data).
Such data could be used to monitor:
- productivity and output in key economic sectors in developing regions such as agriculture, mining, fishing and manufacturing;
- rural infrastructure, such as the prevalence of good roads or advanced healthcare facilities;
- consumer activity, sentiment and trends;
- household consumption and accumulation of wealth; and
- community-level health outcomes, including measures of food security (stunting, wasting), mortality and disease prevalence.
New kinds of data combined with new data science would help to advance scientific understanding of how to most rapidly achieve sustainable prosperity. At the same time, it is likely that new machine- learning approaches for weak supervision and domain adaptation will have to be developed to fully leverage these data sets, perhaps transferring models trained in other, more developed regions with better training data. Thus, this area will require collaboration from experts in various disciplines to fully realize the promise of data science for sustainable development.
Examples of faculty working in the area include David Lobell, Stefano Ermon, Jure Leskovec, Matt Gentzkow and David Grusky.