Forecasting Operating Characteristics of (s, S) Inventory Systems.
Abstract
The report considers the management of an inventory control system operating under incomplete information about the probability distribution of demand. The system control policy is of an (s,S) form where the policy parameters, s and S, are periodically revised using a demand history of fixed length. In such a system it is often desirable to forecast values of such key system operating characteristics as: total cost per period, period-end inventory, period-end backlog quantity, backlog frequency, and replenishment frequency. Several methods of forecasting system performance are presented. The criterion used to evaluate and compare these methods is the extent of prediction bias and its level of dispersion. Most of these procedures require estimation of the underlying demand distribution from a limited sample of historical data. In a 16-item system, forecasts of aggregate total cost are typically 20% below actual values. Much of this error results from underestimation of backlog quantities. This is due to the fact that the demand data is used both to set the policy parameters and to forecast system behavior.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1977
- Accession Number
- ADA049523
Entities
People
- Alastair Maccormick
- Carl R. Schultz
- Richard Ehrhardt
Organizations
- University of North Carolina at Chapel Hill