Optimizing Navy Reservist Assignments

Every three months, roughly 2,000 Navy Reserve Sailors apply to billets. Currently, a time-intensive manual process is used to assign reserve Sailors to billets. This thesis develops a Python-based decision support tool called the Reserve Applied Sailor Model (RASM) to facilitate this process. RASM maximizes number of assignments while considering four goodness-of-fit metrics: unit type, locality, and Sailor and command preference. While teams of assigners currently assign one or a few Sailors at a time, RASM considers all possible assignments and all metrics at once. Each metric is assigned a weight. While there are established default weights, users can input weights, and weights between metrics can vary between assignment iterations based on priorities for each cycle. The model structure is designed to remove subjectivity and bias in assignments and to ensure reproducibility in results. Compared to the manual process, RASM assigns more Sailors, assigns higher percentages of Sailors in favorable metric categories, and completes a three-week assignment task in under two minutes. The time required to input data for RASM, validate its output, and implement the resulting assignment is approximately one week. RASM will optimize fit and fill and will speed up the assignment process within each quarterly cycle, yielding manpower savings. This work will benefit the entire Navy Reserve and will produce tangible increases in lethality and warfighting readiness.