Application of the Bayes Technique to Spare-Parts Demand Prediction

A proposal of an improved procedure for direct use in predicting the demand for spare parts. The study examines a specific aspect of the supply problem, that is, the estimating of spare-parts demand rates at times when data from operational experience are still relatively sparse. While the Air Force has been relying in such instances on a priori estimates made by the initial provisioning conference, shifting to a computed demand rate only at a later time when experience was judged to be "extensive," the study urges a gradual and systematic transition from the one to the other. Even "limited" demand experience is thus used. At any point of time, a priori estimates and observed demands are weighted by expected accuracy. For carrying out this procedure, the study shows how to prepare and use a weighted average of the two sources of data. Graphs, formulae, and a step-by-step procedure for making demand-rate predictions are included.