The Identification and Forecast of Seasonal Demand Consumable Items in Base Supply
Abstract
This study investigated seasonally demanded consumable items at the base-level. This investigation examined how the Standard Base Supply System currently addresses seasonally demanded consumable items and some alternative methods of addressing seasonally demanded consumable items in the SBSS. The alternative methods analyzed in depth were simple and Winters' seasonal exponential smoothing, and Box-Jenkins forecasting models. This study found that items under study display some seasonal demand tendencies. The sample consisted of 12 out of 77 items identified as seasonal by Stock Control personnel at Langley AFB, VA. A graphical analysis showed stronger seasonal demand tendencies than did the autocorrelation functional in which the correlations between demands one year apart are determined. As was expected, the two seasonal models, Winters' exponential smoothing and Box-Jenkins better predicted demands for items under study than the SBSS model. Of 28 forecasts, Winters' exponential smoothing was best 13 times, while Box-Jenkins models were best 9 times. The autocorrelation function could be used to test demand data for seasonality and flag items with seasonal demand patterns for special seasonal treatment, but this is not currently practical. Any useful effort to test all items loaded on the SBSS at a base would require demand data for each item for four or, preferably, more years. Any manual attempts to test items for seasonality would be impractical given the number of items in the average base supply account. Theses.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1989
- Accession Number
- ADA215094
Entities
People
- Mark A. Syzdek
Organizations
- Air Force Institute of Technology