Evaluating and improving the SAMA (Segmentation Analysis and Market Assessment) recruiting model

Abstract

Military recruiting for an all-volunteer force requires deliberate planning and market analysis in order to achieve prescribed recruiting goals. United States Army Recruiting Command (USAREC) leaders and planners leverage existing tools and technology to measure performance and potential within their areas of responsibility. One of the tools used by USAREC is the Segmentation Analysis and Market Assessment (SAMA) tool. This tool calculates recruitment potential of recruiting centers using a four-year weighted performance average within customized Army market segments. An analysis of the current SAMA model shows it overestimates production potential for 96% of centers, leading decision makers to set unrealistic goals for their organizations. The use of additional factors and alternative modeling approaches, Least Squared Regression, and Neural Networks, results in models with greater predictive power, while avoiding overestimation. The statistical models developed in this thesis match the predictive power of the current SAMA methodology while overestimating average potential by only 3.8%. More precise modeling tools will improve USAREC’s ability to more effectively plan recruiting operations and allocate resources.