USING MACHINE LEARNING TO PREDICT EARLY SERVICE SEPARATION OF TECHNICAL AND NON-TECHNICAL SAILORS

Thumbnail Image
Authors
Cole, Stephen
Subjects
machine learning
regression
retention
sailor
technical sailor
non-technical sailor
Advisors
Ahn, Sae Young
Fan, James J.
Date of Issue
2019-12
Date
Publisher
Monterey, CA; Naval Postgraduate School
Language
Abstract
Sailors are difficult to recruit, expensive to train, and hard to retain. This is particularly true in the technical sailor community. Retention of both technical and non-technical sailors is critical to future manning continuity and capability within the Royal Australian Navy. This research employs machine learning to analyze Royal Australian Navy exit survey data collected between 1999 and 2018 to better predict the attitudes and behaviors of a sailor voluntarily separating between four and eight years of service. Furthermore, this study analyzes in particular whether technical sailors behave differently compared to non-technical sailors. In comparison to traditional modeling techniques, the analysis finds that machine learning can more accurately detect differences in the attitudes and behaviors of technical and non-technical sailors when they are deciding to voluntarily separate from service. Furthermore, the analysis can identify differences in sentiment across periods of time covering key career milestones. This analysis and its findings may now be employed to analyze specific critical target groups in both the Royal Australian Navy technical and non-technical sailor communities to understand their attitudes and behaviors, and help support current and future sailor retention policy initiatives.
Type
Thesis
Description
URI
https://hdl.handle.net/10945/64123
Series/Report No
Manpower Systems Analysis Theses
Department
Department of Defense Management (DDM)
Organization
Identifiers
NPS Report Number
Sponsors
Funder
Format
Citation
Distribution Statement
Approved for public release; distribution is unlimited.
Rights
Copyright is reserved by the copyright owner.
Collections
Theses
Load more