Forecast Cloudy: A Case Study in Predicting Conflict Risk
Publisher: New America
Author(s): Peter Kerins and Sharon Burke
Date: 2019
Topics: Assessment, Conflict Causes, Conflict Prevention
Countries: Congo (DRC), Ethiopia, Georgia, Guatemala
Early action requires early warning. Nations and non-governmental organizations alike have long sought the means to anticipate conflict in hopes of preventing violence or mitigating damage. Data analysis techniques, such as mathematical modeling and machine learning, offer new opportunities for forecasting the incidence of organized violence based on underlying conditions, including by using variables that are often absent from more traditional predictions, such as water stress. In December 2019, the Water, Peace, and Security partnership, a group of research organizations sponsored by the Government of the Netherlands, launched a new conflict prediction model using these advanced techniques. This case study peels back the curtain to show how the research team built the model, how decision makers might use such a tool, and observations from a "ground truthing" trip the team took to an area in Ethiopia identified by the model as at high risk for conflict.