NICP FORECASTING TECHNIQUES.
Abstract
This report contains recommendations on methodology of forecasting demands and returns of spare parts and components. These recommendations resulted from mostly empirical studies done by the Inventory Research Office (U. S. Army Materiel Command). The studies and analyses leading to these recommendations are described in detail. Included are time series analysis, methods of forecasting, and study of outlier demands. The most appropriate techniques for the three major aspects of forecasting - forecasting the mean, the variability, and detection of significant change in the underlying stochastic process - are discussed and evaluated. In forecasting the mean, these include simple moving average and single and double exponential smoothing. A range of bases is investigated for best results. Also investigated is the effect of non-recurring demands on the accuracy of the forecasted mean. In forecasting the variability the techniques considered are those dependent on the availability of past history (ordinary statistical estimates) and those independent of the availability of past history (using a nomograph). The feasibility of the latter is investigated. Tracking signals of R. G. Brown and D. W. Trigg are discussed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1967
- Accession Number
- AD0822934
Entities
People
- S. Gajalo