The Statistical Foundations of the Pilot Prediction System
Abstract
The Pilot Prediction System (PPS) is a research effort designed to provide Navy policy makers with improved access to selection and training data in the aviation community. One of its main features is the ability to make predictions about the future success of aviation candidates in flight training. The purpose of this report is to present in some detail the statistical foundations of this feature of the PPS. We first describe the rudiments of statistical decision theory. Such a theory allows us to make the very best decision possible when faced with the inherent uncertainty about what is actually going to take place in the future. The second pillar on which the PPS is built is the treatment of probability from the Bayesian perspective. Two technical appendices are included for the interested reader. The first contains a simplified proof of the Bayesian predictive density that allows the prediction algorithm in the PPS to be written in its most general form. The second shows an analytical solution derived from the theory developed in the first appendix justifying a practical approximation for predicting pass or fail during flight training.
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 27, 2001
- Accession Number
- ADA529804
Entities
People
- D. J. Blower
Organizations
- Naval Aerospace Medical Research Laboratory