A Comparison of Demand Forecasting Techniques

For periodic review inventory models with stochastic demand, the idea of stock-out risk is defined, from which the importance of accurate prediction of demand is deduced. Methods of demand model parameter estimation are investigated and several methods compared on the basis of theoretical soundness, ease of application, and accuracy of estimates based upon the results of extensive computer simulation. The theoretical development of maximum likelihood and exponential smoothing estimators as applied to prediction is presented along with the development of a new Bayesian approach to the problem of demand forecasting.