基于支持向量机方法的金融时间序列研究  

Financial Time Series Research Based on Support Vector Machine's Means

作　　者：张立霞[1] 马芳芳[2] 叶德谦[1] 

机构地区：[1]辽宁工业大学信息科学与工程学院,辽宁锦州121001 [2]秦皇岛职业技术学院信息工程系,河北秦皇岛066100

出　　处：《辽宁工业大学学报（自然科学版）》2008年第1期27-30,共4页Journal of Liaoning University of Technology(Natural Science Edition)

基　　金：教育部留学回国人员科研基金(2001498)

摘　　要：支持向量机是基于统计学的一种新型的机器学习和数据挖掘的技术,实现了结构风险最小化原则。由于金融时间序列是非平稳的、复杂的,非线性的,含有噪声数据,传统的方法很难得到满意的预测效果。提出了基于支持向量机的金融时间序列预测的方法,应用到我国上证180指数预测中,实验结果表明支持向量机方法对动态的金融时间序列具有较好的建模能力,达到了较好的预测效果。Support vector machine was one kind of the late-model technology in machine learning and data mining, which was based on statistics, has realized structural risk minimization principal. Because financial time series was unstable, complicated, nonlinear, and containing noise data, it was very difficult to get the satisfied forecasting effect. A method forecasting financial time series based on support vector machine was proposed in the research, which was applied in predicting 180 Shanghai Stock Index of our China. The experiment expatiated that support vector machine method has fairly good modeling capability about dynamic financial time series, and reached fairly good forecasting effect.

关 键 词：支持向量机 数据挖掘 结构风险 金融时间序列 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]