基于漂移度的组合预测方法研究

A Combination Forecasting Model based on Drift

由于不同的预测方法能够提供不同的有用信息,其预测精度往往也存在差异,为了分散预测的风险,采用组合预测方法。本文首先提出相容方法集和互补模型集,然后在对不同单一预测模型的漂移性和互补性研究的基础上提出了基于漂移度的组合预测模型,为组合预测模型研究提供一种新的思路。最后通过实例来说明基于漂移度的组合预测模型能够提高样本期预测精度和外推预测精度及实际应用的有效性。

As different methods diverse in providing information and forecast precision,the combined forecasting model is employed.Combination forecasting model based on drift is proposed on basis of the study on drift and complementarity of different single forecasting models.A new thought of combination forecasting model is given.Finally some practical examples are given to show that combination forecasting model based on drift can improve forecasting precision and is effective in practice.