The last decade has seen an explosion in the capabilities of quantitative forecast methods for armed conflict prediction. These are currently widely deployed for analytical purposes. However, their use for prevention poses challenges as policy options are typically not evaluated in systems that integrate quantitative forecasts.
The Dynamic Early Warning and Action Model (DEWAM) is an attempt to fill this gap. It combines two modules: 1) a forecasting module which uses machine learning and millions of news articles to predict the outbreak and intensity of armed conflict twelve months before they occurr. 2) a decision-making module which uses dynamic programming, a technique often used in economics to solve complex dynamic decision problems. The result is a list of countries which we also publish on the webpage conflictforecast.org.
This project is an attempt to bring this complex project to a broader audience by explaining the different parts of the model. We pay special attention to explaining the risk stages/states model to an audience that does not know what a Hidden Markov Model is :-)
The report can be downloaded here. The online version and the latest prevention gains update can be seen here.