Abstract:

In order to enhance the forecasting accuracy of materials with intermittent demand, a combined forecasting model is proposed. This model disintegrates the time series into explanatory series and auto-correlated series. Then the probabilities of nonzero demands for the two series in lead time are estimated by Logistic regression model and Markov forecasting respectively. The final demand forecasting result is calculated through integrating the results of the two series. Experiment results show that the forecasting model can get high precision.