Executive Summaries

摘要

Basic methods for determining which historical observations to use in modeling counterinsurgencies are not well documented. The authors develop an approach that addresses this problem by evaluating existing counterinsurgency classification schemes to determine which historical observations to use for predictive analysis. They find that the insurgent strategy classification scheme, when used to estimate annual counterinsurgent deaths, produces the lowest prediction error. The authors' results support the application areas of modeling, simulation, and wargaming, using methods from applied statistics including categorical data analysis and forecasting.