Bayesian Approach to Modelling of Quasi-Periodic Intermittent Demand
| Authors: | Dolgui Alexandre, Ecole des Mines de Saint Etienne, France
Pashkevich Anatoly, Belarusian State University of Informatics and Radioelectronics, Belarus
Pashkevich Maxim, Belarusian State University, Belarus |
|---|
| Topic: | 5.1 Manufacturing Plant Control |
|---|
| Session: | Plant-wide Production Planning and Control Issues |
|---|
| Keywords: | inventory control, demand modelling, statistical inference, parameterestimation, prediction methods. |
|---|
Abstract
The paper focuses on the stochastic modelling of the quasi-periodic intermittentdemand patterns, which arise in the inventory management of the “slowmoving items” such as service parts or high-priced capital goods. It is proposed anew stochastic model, which describes the demand patterns with essentially nonexponentialdistribution of the inter-arrival times. The model is based on generalizedbeta-binomial distribution and the Bayesian inference using the historical data arraydescribing the demand repeatability within the time periods. For this model, therewere derived explicit expressions for the forecast distributions, its moments andrelevant Bayesian risk. The efficiency of the proposed approach is confirmed bycomputer simulation and an application example.
