【作者】 董景荣；

【Author】 Dong Jingrong(School of Business Administration,Chongqing University,Chongqing 400044,China)

【机构】 重庆大学工商管理学院博士后流动站；

【摘要】 针对汇率变化的非线性特性与线性组合预测方法在解决非平衡时间序列的组合建模问题所遇到的困难和存在的不足,本文建立了汇率的非线性组合预测模型,并确定了有效的学习算法。应用从1985年1月至1992年12月英镑对美元的月末数据,进行了实证研究。结果表明:该方法具有很强的学习与泛化能力,在处理诸如汇率时间序列这种具有一定程度不确性的非线性系统的组合建模与预测方面有很好的应用价值。更多还原

【Abstract】 :Taking aim at the nonlinear nature of the exchange rate data and the problem how to overcome thedifficulties and drawbacks in combined modeling non-stationary time series by using linear combinationforecasting.A nonlinear combination forecasting model for exchange rate is presented and a effective learningalgorithm is also produced.A experiments research on combined modeling and forecasting the exchange ratesofthe British pound/US dollar over the period from January 1985 to December 1992 monthly show that thenew techniques has reinforcement learning properties and universalized capabilities.With respect tO combinedmodeling and forecasting the dynamics of exchange rate series in nonlinear systems,which have someuncertainties.the method iS feasible and effective.更多还原