PREDICTIVE TECHNIQUES STUDY. PHASE I. COMPARISON OF SOME FORECAST METHODS.
Operations analysis rept.,
AIR FORCE LOGISTICS COMMAND WRIGHT-PATTERSON AFB OHIO OPERATIONS ANALYSIS OFFICE
Pagination or Media Count:
The report discusses the results of the research performed on Phase I of the Predictive Techniques Study. Phase I consisted of a short-term effort to obtain indications whether single or double exponential smoothing, double moving average, or moving least squares are more accurate than single moving average for forecasting. Phase I dealt only with computer generated artificial demand patterns thirty-nine runs were made on various demand patterns, and the alternative forecast methods evaluated for each run. A unique feature is that varying forecast lead times are taken into account short-term vs. medium-term vs. long-term. Other parameters being varied include the exponential smoothing constant alpha, mean demand, variance of demand random fluctuation, slope of trend when applicable. Author
- Administration and Management
- Logistics, Military Facilities and Supplies