基于组合预测的间断性需求器材预测

Forecasting for materials with intermittent demand based on combined forecasting

为了提高间断性需求装备器材的预测精度,提出一种组合预测模型。该模型从解释变量序列和自相关序列两个方面进行组合预测,对解释变量序列采用Logistic回归模型预测提前期非零需求发生概率,对自相关序列采用Markov过程估计提前期非零需求发生概率,整合两个预测结果得到最终的提前期需求。实验结果表明,该预测模型具有较高的预测精度。

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.