Improved Forecasting Methods for Naval Manpower Studies
Abstract
Forecasted manpower inventory, the number of individuals available in a given time period, are derived from stay/loss models, where estimates of the probability of staying in the navy informs the advancement and gains modules used within the Department of the Navy. As such, the accuracy of these probability rates is critical to these related functions. Extending an earlier study, we focus on two methodologies, autoregressive and logistic methods, and consider the effect of structural changes on forecast accuracy. Exogenous events or structural breaks in time-series data can result in large forecasting errors. Using the Bai-Perron (BP) test, we determine if structural breaks occur in the data. In cases where breaks are identified we control for breaks in the models. We then validate and discuss improvements in the forecast accuracy.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 25, 2015
- Accession Number
- ADA614084
Entities
People
- Ping Y. Bellamy
- Tanja F. Blackstone
Organizations
- Navy Personnel Research, Studies, and Technology