摘要
本文利用一个新的非参数支持向量回归(SVR)方法来预测基于非线性ARI模型的汇率时序变量,并且与最大似然法(MLE)和人工神经网络(ANN)的预测结果进行比较。从理论上讲,MLE和ANN方法仅侧重于样本内拟合,而SVR方法则同时考虑了拟合和预测,因此,其预测能力在现有方法中是最强大的。本文选择中国、韩国、印度和瑞士四种货币的日汇率来进行预测检验,实证结果支持SVR方法具有最强的预测能力。
This paper applies Support Vector Regression （SVR） to forecast ARI model of the daily exchange rates of four currencies （Swiss Francs, Indian Rupees, South Korean Won and China Renminbi） . The forecasting abilities of SVR are compared with MLE and ANN approaches. Theorically, MLE and ANN only focus on fit within-sample, but SVR considers both fit and forecast out-of-sample. Therefore SVR has the strongest forecast ability among all methods. Empirical findings are in favor of SVR.
出处
《数量经济技术经济研究》
CSSCI
北大核心
2007年第5期142-150,共9页
Journal of Quantitative & Technological Economics
关键词
支持向量回归
ARI模型
汇率预测
预测精度
嵌套检验
Support Vector Regressionl ARI Model
Exchange Rate Forecas-ting
Forecasting Accuracy
Encompassing Test
