An Exploratory Analysis of Time Series Econometric Data for Retention Forecasting using Deep Learning
Date of Award
Master of Science
Department of Operational Sciences
Raymond R. Hill, PhD
Officer retention in the Air Force has been researched many times in an attempt to better predict the personnel needs of the Air Force for the future. There has been previous work done in regards to specific AFSCs and how their retention compares to specific yet similar private sector jobs. This study considers different econometric time series statistics as a feature space and an average Air Force officer separation rate as the response variable for the multivariate time series analysis deep learning techniques. The econometric indicators used in this study are New Business Formations, New Durable Good Orders, and the Consumer Confidence Index. The techniques considered for this study were Long Term Short Memory (LSTMs) Networks and Gated Recurrent Unit (GRU) Networks. This study shows that both GRUs and LSTMs perform fairly well with a forecast of 14 months out, but does not perform well comparatively to the more traditional univariate time series forecasting techniques, ARIMA models. The career fields with better performing models were career fields that will have jobs outside of the Air Force that will be more likely to hire in a period of economic growth, which would in turn increase the separation rate.
DTIC Accession Number
O'Donnell, John C., "An Exploratory Analysis of Time Series Econometric Data for Retention Forecasting using Deep Learning" (2022). Theses and Dissertations. 5351.