Improving International Development Through Data-Driven Methods

The Peace Corps is an independent agency run by the U.S. Federal Government that primarily sends American volunteers abroad for two years programs to complete international development projects. However, 10-15% of the time, Peace Corps Volunteers do not complete their two years of service when they resign due to various factors such as inability to integrate into the community, lack of access of certain resources, or difficulties completing their projects. When this happens, communities are often left with unfulfilled expectations which also diminishes the Peace Corps’ credibility in international development. This summer, I explored this issue while working at the agency’s Office of Strategic Information, Research, and Planning (OSIRP) by using existing data from Volunteers to create a predictive machine learning model of resignations. Inputting a Volunteer’s annual survey responses which includes data ranging from demographics to satisfaction with service to living conditions, the model can predict with 65-70% accuracy whether that Volunteer will resign. The hope is that this project can act as a first step in informing actionable—and data evidenced—policies within the agency to address the issue of Volunteer resignations and improve its impact in international development.