Recognizing Patterns of Nation-State Instability that Lead to Conflict

This paper presents research conducted at the Center for Army Analysis (CAA) to classify patterns of nation-state instability that lead to conflict Nation-states that experience instability-induced conflict are similar with respect to several factors and patterns We utilize standard pattern classification algorithms (k-Nearest Neighbor and Nearest Centroid) to identify these patterns First, we identify features (political, military, economic and social) that capture the instability of a nation-state Second, we forecast the future levels of these features for each nation-state Third, we classify each future state's conflict potential based upon the conflict level of those states in the past most similar to the future state This method improved upon the accuracy of earlier predictive models at CAA and provided an understandable geometric interpretation of the results for senior decision makers in the Army