基于ARIMA与GRNN组合模型对人民币汇率的预测
发布时间：2018-10-05 13:25
【摘要】：随着中国经济的高速发展和汇率市场化进程的不断加快,人民币汇率在国内外经济中的地位越来越重要。2015年8月11日,为增强人民币汇率的市场化程度,让人民币顺利加入特别提款权,央行放弃了对每日人民币兑美元中间价的干预,汇率波动幅度越来越大,短期汇率走势也具有很强的不确定性。在此背景下,准确预测汇率波动性对降低中国企业的生产成本和规避汇率风险具有至关重要的理论研究意义和实际应用价值。本文在国内外学者对汇率的研究基础上,根据汇率序列具有线性和非线性的复合特征,并利用ARIMA模型和GRNN模型分别在线性空间和非线性空间的预测优势,并构建了ARIMA与GRNN的组合模型对人民币汇率波动趋势进行实证分析。本文主要分为五部分:第一部分首先对人民币汇率的研究背景及意义进行了阐述,强调该课题研究的重要性,然后对人民币汇率国内外的研究动态进行了概述,主要包括国内外学者对汇率波动预测的时间序列分析法研究以及文献评述,最后概括性地阐述了本文研究的主要内容和方法。第二部分回顾了人民币汇率制度的发展过程及不同制度下的人民币兑美元汇率的变化趋势,并分析了人民币兑美元汇率波动给我们日常生活及国家经济带来的影响。第三部分介绍了ARIMA模型和GRNN模型的相关理论以及模型参数设置的方法,紧接着阐述了ARIMA与GRNN的组合模型的原理及建模步骤。第四部分利用ARIMA与GRNN组合模型对人民币汇率进行实证分析。首先利用ARIMA模型对人民币兑美元汇率中间序列进行预测分析得到线性主体部分;然后利用GRNN神经网络模型对前一模型的残差进行预测分析得到非线性部分;最后将线性主体部分与非线性残差部分相加得到人民币兑美元汇率中间价序列的预测结果。研究结果表明,ARIMA-GRNN组合模型对人民币兑美元汇率中间价的预测效果优于单一的ARIMA模型和单一的GRNN模型,且GRNN模型比ARIMA模型的预测效果好。第五部分在对论文实证检验的结果进行分析总结的基础上,还指出了论文的不足之处以及需要进一步值得研究的地方。
[Abstract]:With the rapid development of Chinese economy and the accelerating process of marketization of exchange rate, RMB exchange rate is becoming more and more important in domestic and foreign economy. On August 11, 2015, to enhance the marketization of RMB exchange rate, Allowing the yuan to join the SDR smoothly, the central bank dropped its daily intervention against the dollar, the exchange rate fluctuated more and more, and short-term exchange rate movements were highly uncertain. Under this background, accurate prediction of exchange rate volatility is of great theoretical significance and practical application value in reducing production cost and avoiding exchange rate risk of Chinese enterprises. Based on the domestic and foreign scholars' research on exchange rate, according to the linear and nonlinear characteristics of exchange rate series, this paper makes use of the prediction advantages of ARIMA model and GRNN model in linear space and nonlinear space, respectively. The combination model of ARIMA and GRNN is constructed to analyze the fluctuation trend of RMB exchange rate. This paper is divided into five parts: the first part expounds the research background and significance of RMB exchange rate, emphasizes the importance of the research, and then summarizes the domestic and foreign research trends of RMB exchange rate. This paper mainly includes the research of time series analysis and literature review on the prediction of exchange rate fluctuation. Finally, the main contents and methods of this paper are summarized. The second part reviews the development process of RMB exchange rate system and the change trend of RMB / US dollar exchange rate under different systems, and analyzes the impact of RMB / US dollar exchange rate fluctuation on our daily life and national economy. The third part introduces the related theories of ARIMA model and GRNN model and the method of model parameter setting. Then the principle and modeling steps of the combined model of ARIMA and GRNN are expounded. The fourth part uses ARIMA and GRNN combination model to analyze RMB exchange rate empirically. Firstly, the linear principal part is obtained by using the ARIMA model to predict the intermediate sequence of RMB / US dollar exchange rate, and then the nonlinear part is obtained by using the GRNN neural network model to predict the residual of the former model. Finally, by adding the linear principal part and the nonlinear residual part, the prediction results of the intermediate price series of RMB / US dollar exchange rate are obtained. The results show that the ARIMA-GRNN combination model is superior to the single ARIMA model and the single GRNN model in predicting the intermediate price of RMB / US dollar exchange rate, and the GRNN model is better than the ARIMA model. On the basis of analyzing and summarizing the results of empirical test, the fifth part also points out the shortcomings of the paper and the points that need to be further studied.
【学位授予单位】：湖北工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.6
本文编号：2253554
[Abstract]:With the rapid development of Chinese economy and the accelerating process of marketization of exchange rate, RMB exchange rate is becoming more and more important in domestic and foreign economy. On August 11, 2015, to enhance the marketization of RMB exchange rate, Allowing the yuan to join the SDR smoothly, the central bank dropped its daily intervention against the dollar, the exchange rate fluctuated more and more, and short-term exchange rate movements were highly uncertain. Under this background, accurate prediction of exchange rate volatility is of great theoretical significance and practical application value in reducing production cost and avoiding exchange rate risk of Chinese enterprises. Based on the domestic and foreign scholars' research on exchange rate, according to the linear and nonlinear characteristics of exchange rate series, this paper makes use of the prediction advantages of ARIMA model and GRNN model in linear space and nonlinear space, respectively. The combination model of ARIMA and GRNN is constructed to analyze the fluctuation trend of RMB exchange rate. This paper is divided into five parts: the first part expounds the research background and significance of RMB exchange rate, emphasizes the importance of the research, and then summarizes the domestic and foreign research trends of RMB exchange rate. This paper mainly includes the research of time series analysis and literature review on the prediction of exchange rate fluctuation. Finally, the main contents and methods of this paper are summarized. The second part reviews the development process of RMB exchange rate system and the change trend of RMB / US dollar exchange rate under different systems, and analyzes the impact of RMB / US dollar exchange rate fluctuation on our daily life and national economy. The third part introduces the related theories of ARIMA model and GRNN model and the method of model parameter setting. Then the principle and modeling steps of the combined model of ARIMA and GRNN are expounded. The fourth part uses ARIMA and GRNN combination model to analyze RMB exchange rate empirically. Firstly, the linear principal part is obtained by using the ARIMA model to predict the intermediate sequence of RMB / US dollar exchange rate, and then the nonlinear part is obtained by using the GRNN neural network model to predict the residual of the former model. Finally, by adding the linear principal part and the nonlinear residual part, the prediction results of the intermediate price series of RMB / US dollar exchange rate are obtained. The results show that the ARIMA-GRNN combination model is superior to the single ARIMA model and the single GRNN model in predicting the intermediate price of RMB / US dollar exchange rate, and the GRNN model is better than the ARIMA model. On the basis of analyzing and summarizing the results of empirical test, the fifth part also points out the shortcomings of the paper and the points that need to be further studied.
【学位授予单位】：湖北工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.6
【参考文献】
相关期刊论文 前10条
1 张艳;蔡光兴;;基于ARIMA和GRNN模型对人民币汇率的预测[J];特区经济;2017年02期
2 王正新;陈雁南;;“汇改”以来人民币实际有效汇率波动的非线性机制——基于平滑转移自回归模型的实证研究[J];金融理论与实践;2016年06期
3 孔佳文;卞佳yN;方小萱;李文;;基于ARIMA模型的人民币汇率分析及预测[J];现代经济信息;2016年11期
4 孙永利;王华金;郝丽;肖晓明;;基于神经网络和遗传算法的螺旋折流板换热器性能预测[J];化学工业与工程;2016年04期
5 宫舒文;;基于GARCH族模型的人民币汇率波动性分析[J];统计与决策;2015年12期
6 王建国;;基于BP神经网络的股票价格反转点预测[J];现代计算机(专业版);2015年05期
7 张莹;邵毅;王式功;尚可政;李旭;刘慧;耿迪;;呼吸系统疾病死亡人数的人工神经网络方法研究[J];中国卫生统计;2014年05期
8 李明景;汪金菊;;基于ARMA-稀疏贝叶斯模型的汇率预测研究[J];合肥工业大学学报(自然科学版);2014年08期
9 危黎黎;李超;李余辉;;基于STAR模型的人民币汇率非线性特征及预测[J];统计与决策;2014年09期
10 陈黎明;王春香;黄伟;胡晋武;;人民币汇率波动的非线性特征研究[J];统计与决策;2014年09期
相关博士学位论文 前1条
1 廖薇;基于神经网络和遗传规划的汇率预测技术研究[D];华东师范大学;2010年
相关硕士学位论文 前2条
1 陈天舒;基于ARIMA与GPR组合模型的人民币汇率预测[D];山东大学;2015年
2 黄腊梅;人民币汇率GARCH-GRNN组合预测研究[D];湖南大学;2009年
，本文编号：2253554
本文链接：https://www.wllwen.com/jingjilunwen/huobiyinxinglunwen/2253554.html
