Please use this identifier to cite or link to this item: https://dair.nps.edu/handle/123456789/4713
Title: Machine Learning in AWF Talent Management: New Approaches to Prediction of Workforce Retention and Promotion
Authors: Tom Ahn, James Fan
Keywords: acquisition workforce
retention
AWF
machine learning
talent management
Issue Date: 24-Jun-2022
Publisher: Acquisition Research Program
Citation: APA
Series/Report no.: Human Resources;NPS-HR-22-192
Abstract: The Department of Navy (DoN) and Department of Defense (DoD) Acquisition Workforce (AWF) Strategic Plans call for a restoration and strengthening of the civilian AWF after more than two decades to contraction. To reform and reshape the workforce to improve the acquisition process and delivery of world-class warfighting capability for the military, the AWF leadership must understand how attrition and retention will impact the “size, composition, and skill” needs of the workforce in “parallel with technology advances and global trends.” To achieve strategic talent management of the workforce, it is critical to have the ability to predict which workers are most likely to leave the AWF. Forecasting attrition will aid the leadership by identifying 1) which workers to target for retention via incentives and 2) which areas will need to increase or decrease recruitment to quickly fill personnel gaps that may arise. This technical report is the first to evaluate whether Machine Learning (ML) can be a useful tool for the AWF leadership to make attrition forecasts. We first show that ordinary least squares (OLS) which is the tool most-often associated with statistical modeling of worker behavior performs poorly, especially given the sparse administrative dataset we have access to. We then test a variety of ML algorithms and find that they can predict worker attrition with a higher degree of accuracy. Our conclusion from this exploratory analysis is that, as algorithmic effective increases with dataset size (in terms of more worker and job/task characteristics), there may be many use cases for these algorithms in future predictive modeling for manpower and retention.
Description: Faculty Report
URI: https://dair.nps.edu/handle/123456789/4713
Appears in Collections:Sponsored Acquisition Research & Technical Reports

Files in This Item:
File Description SizeFormat 
NPS-HR-22-192.pdfTechnical Report1.11 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.