Adaptive Forecasting of the Size of a Force Subject to Random Withdrawals,
Abstract
Adaptive forecasting procedures are developed for a problem of predicting the size of a force which is subject to random withdrawals. The structure of the problem is phrased in terms of the Marine Corps. The author distinguishes between four phases in the service of incoming cohorts. Methods of forecasting have been developed for each phase separately. The present study concentrates on Phase 2 forecasting. The author compares three forecasting procedures: conditional maximum likelihood estimation of prediction intervals; tolerance intervals, and Bayes prediction intervals. The theory of these forecasting procedures is developed and numerical examples illustrate it.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 09, 1975
- Accession Number
- ADA013175
Entities
People
- S. Zacks
Organizations
- George Washington University