摘要
在股市投资测试问题的研究中,股价是一种高度不稳定、复杂且难以预测的时间序列数据,传统预测方法都是基于线性模型,忽略了股价的非线性特征,导致预测精度不高。为解决股价预测过程中的精度不高的难题,提出支持向量机引入到股价预测的建模中。首先采用支持向量机非线性扩展样本对时间序列模型定阶,并利用前向浮动特征筛选法选择特征,建立基于支持向量机的股市预测系统模型,对股价进行仿真实验。仿真结果表明,支持向量机模型比神经网络和CAR模型有较高的预测精度,证明适用于股市预测等非线性问题的预测,且有较高的精确度和应用价值。
Stock investment has become an important part of people＇s daily life. Forecasting the stock price has been a concern problem. Stock market is a complicated nonlinear dynamic system. It is very difficult to open out its inherent rulesusing traditional timing prediction technique. To improve the analysis of the pattern of the stock market price,based on the theory of support vector machines,we put forward a new method for forecasting stock price. Firstly,the time series model-order is determined by non-linear expansion of samples with SVM,then a floating search method for factor selection is proposed based on support vector regression and leaving-one method.The test shows that the accuracy of the prediction is obviously higher than traditional NN classification ways,such as BP slgorithm and thus it has a satisfying result.
出处
《计算机仿真》
CSCD
北大核心
2010年第9期302-305,共4页
Computer Simulation
关键词
股价预测
支持向量机
预测
留一法
Stock price prediction
Support vector machine（SVM）
Prediction
Leaving-one method
