Predicting U.S. Army First Term Attrition After Initial Entry Training

The United States Army recently announced a reduction of its 2018 recruiting goal due to a challenging recruiting environment and a shrinking population of eligible candidates. However, the Sergeant Major of the Army has stated that the current improvement in the retention of existing soldiers should mitigate the loss of new recruits. The goal of this research is to identify demographic and administrative factors of active-component, first-term enlisted soldiers who have completed their Initial Entry Training to construct predictive models capable of identifying soldiers with high chances of failure in completing their initial contractual obligation. We construct a binary logistic regression model and a random forest classification model to predict a soldier’s probability of first-term attrition based on the individual’s unique service record. We find that a soldier’s deployment history and the duration of the initial contract are significant predictors of whether a soldier will complete his or her first term. Knowledge of the key factors, and other influencing variables, assists the Army Resiliency Directorate in creation of models and tools to better advise Army leadership and development of intervention strategies and preventative measures to prevent the loss of first-term soldiers.