预测组合权重更新的过程控制及分析

发布时间：2018-05-23 16:18

本文选题：预测组合 + 赋权方法　；参考：《数理统计与管理》2017年06期

【摘要】：源于与决策分析的相关性,预测组合已经逐渐形成了一个重要的研究领域。为此,本文引进EWMA技术对预测组合权重更新的过程进行控制,从而提出一种能够应用于实际且简单有效的EWMA赋权方法。这种赋权方法能够确定预测组合权重应该何时更新,而不是机械地更新预测组合权重。本文额外针对各种赋权方法在旅游预测组合模型中的预测性能(全面预测性能和总均方根误差)和预测效率(权重更新频率)进行了经验评估。结果显示:EWMA赋权方法的预测性能优于传统的赋权方法,并与CUSUM赋权方法相似,同时该赋权方法获得了最小的权重更新频率。综合考虑预测性能和预测效率,EWMA赋权方法相比于其他赋权方法在旅游实际应用过程中更具优势。
[Abstract]:From the correlation with decision-making analysis, predictive combination has gradually formed an important research field. Therefore, this paper introduces EWMA technology to control the updating process of predictive combined weights, and proposes a simple and effective EWMA weighting method which can be applied to practice. This method can determine when the predictive weight should be updated, not mechanically. In this paper, the prediction performance (total prediction performance and total root mean square error) and prediction efficiency (weight renewal frequency) of various weighting methods in the tourism forecasting combination model are evaluated empirically. The results show that the prediction performance of the weight weighting method is superior to that of the traditional weighting method, and is similar to that of the CUSUM weighting method, and the weight updating frequency of the weighting method is minimum. The EWMA weighting method is more advantageous than other weighting methods in tourism application considering the prediction performance and efficiency.
【作者单位】：东南大学经济管理学院;四川师范大学数学与软件科学学院;
【基金】：国家自然科学基金面上项目(11371267) 国家社会科学基金青年项目(16CGL026)资助
【分类号】：F224;F592

【相似文献】