基于ARMA-稀疏贝叶斯模型的汇率预测研究  

Exchange rate forecast using ARMA-sparse Bayesian model

作　　者：李明景[1] 汪金菊[1] 

机构地区：[1]合肥工业大学数学学院,安徽合肥230009

出　　处：《合肥工业大学学报（自然科学版）》2014年第8期1019-1024,共6页Journal of Hefei University of Technology：Natural Science

基　　金：安徽省自然科学基金资助项目(1208085MF91;11040606M03);教育部人文社会科学研究基金资助项目(10YJA910005)

摘　　要：文章基于ARMA模型和稀疏贝叶斯模型,提出了ARMA-稀疏贝叶斯模型,充分利用ARMA模型和稀疏贝叶斯模型在线性模型及非线性模型预测中的优势,将收益率序列分解为线性自相关主体和非线性残差2个部分,然后用ARMA模型对线性自相关主体数据进行预测估计,用稀疏贝叶斯模型对非线性残差进行预测估计,最后合成人民币兑美元日汇率中间价序列预测结果。研究结果证明,运用所建模型预测人民币日汇率中间价和上证指数收盘价,均取得了较好的效果。The ARMA-sparse Bayesian model is put forward based on ARMA model and sparse Bayes-ian model .The model takes full advantage of the unique character of ARMA and sparse Bayesian models in linear and nonlinear modeling .T he yield sequence is decomposed into the linear component and nonlinear residual component .And then the ARMA model is used to forecast the linear compo-nent ,w hile the sparse Bayesian model is used to forecast the nonlinear residual component .Finally , the relevant result is forecasted from the synthesis of the RMB against the US dollar rate sequence . The experimental results show that the forecast of RMB exchange rate and the closing price of SSE Composite Index using ARMA-sparse Bayesian model achieves good results .

关 键 词：ARMA模型 稀疏贝叶斯模型 ARMA-稀疏贝叶斯模型 汇率 预测 

分 类 号：O212.8[理学—概率论与数理统计]