Predicting Individual USNR Enlisted Attrition
Abstract
U.S. Navy Reserve sailors are trained to conduct critical operational missions and support the Active-Duty component. They also manage the administration and training of the Reserve program. Despite the importance of these personnel, in many recent years end-strength levels have not been met. This problem has arisen because the current end strength model has not accurately predicted these shortfalls. The variability in the accuracy of the attrition prediction input, a four-year weighted average, presents the difficulty of predicting Reserve attrition. While this thesis does not aim to replace the current aggregate model, it does aim to forecast individual attrition by using medical, administrative, and demographic factors to fit binary logistic regression models that predict whether a service member will attrit in the following year. This study differs from other individual attrition models in that they focus solely on first-term and early attrition that directly impacts recruiting. The results show that improvements to the model are required to increase accuracy. Inclusion of medical variables, as seen in prior theses, and inclusion of Navy Reserve specific variables may be beneficial to identify a subset of variables that can improve the model's predictive power.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2022
- Accession Number
- AD1201696
Entities
People
- Bria N Rand
Organizations
- Naval Postgraduate School