| 基于SVR的金融时间序列预测 |
| |
| 引用本文： | 李立辉,田翔,杨海东,胡月明.基于SVR的金融时间序列预测[J].计算机工程与应用,2005,41(30):221-224. |
| |
| 作者姓名： | 李立辉 田翔 杨海东 胡月明 |
| |
| 作者单位： | [1]湖南师范大学商学院,长沙410081 [2]华南理工大学应用数学系,广州510640 [3]华南农业大学信息学院,广州510642 |
| |
| 摘 要： | 介绍了支持向量回归的建模原理及常用版本,详细探讨了利用支持向量回归方法建立金融时间序列预测模型,进行单步预测和多步预测的步骤。将它们应用到我国上证180指数预测中,并且比较了它们的预测性能。数值实验表明,SVR方法对非平稳的金融时间序列具有良好的建模和泛化能力。特别是LS-SVR用等式约束代替传统支持向量机中不等式约束,使求解过程从解QP问题变成解一组等式方程,因此学习速度更快,并具有更好的预测效果。
|
| 关 键 词： | 人工智能 预测模型 支持向量回归 金融时间序列 非线性建模 |
| 文章编号： | 1002-8331-（2005）30-0221-04 |
| 收稿时间： | 2005-01 |
| 修稿时间： | 2005-01 |
Financial Time Series Forecasting Based on SVR |
| |
| Li Lihui, Tian Xiang, Yang Haidong, Hu Yuemin.Financial Time Series Forecasting Based on SVR[J].Computer Engineering and Applications,2005,41(30):221-224. |
| |
| Authors: | Li Lihui Tian Xiang Yang Haidong Hu Yuemin |
| |
| Affiliation: | 1.College of Business,Hunan Normal University,Changsha 410081; 2.Dept. of Applied Mathematics, South China Univ.of Tech., Guangzhou 510640; 3.College of Imformation,South China Univ. of Agricuhure,Guangzhou 510642 |
| |
| Abstract: | The modeling principle of using support vector regression(SVR)and several popular SVR versions are introduced.A modeling method for financial time series based on SVR is proposed and the detail steps are given.Applying the method to 180-stock-index of Shanghai,some models are built and their forecasting performances are compared.Numerical test results show that SVR has good ability of modeling nonstationary financial time series and good generalization under small data set available.Especially,the LS-SVR solution follows directly from solving a set of linear equations instead of quadratic programming,therefore it can be trained faster and have more satisfactory performance. |
| |
| Keywords: | artificial intelligence forecasting model support vector regression financial time series nonlinear modeling |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
