基于ARIMA-SVM组合模型的股票价格预测  

Research on Stock Price Prediction Based on Wavelet Decomposition and ARIMA-SVM Combined Model

作　　者：程昌品[1] 陈强[1] 姜永生[1] 

机构地区：[1]广东第二师范学院计算机科学系,广东广州510303

出　　处：《计算机仿真》2012年第6期343-346,共4页Computer Simulation

基　　金：广东省科技计划资助项目(2010B010600018);2011年广东省现代信息服务业发展专项资金项目(13090)

摘　　要：现有的股票价格准确预测方法各有优缺点,为了发挥各种预测方法的优点,提出二进正交小波变换和ARIMA-SVM方法的非平稳时间序列预测方案。使用小波分解算法对数据进行分解,分离出非平稳时间序列中的低频信息和高频信息;然后对高频信息构建自回归模型ARIMA预测,对低频信息则用SVM模型进行拟合;最后将各模型的预测结果进行叠加,从而得到原始时间序列的预测值。将预测结果与实际值比较,组合模型具有较好的预测效果。经实验证明,小波分解的ARI-MA-SVM组合模型较单一的预测模型效果更为理想。Existing stock forecasting methods have advantages and disadvantages. In order to make full uses of the advantages of stock forecasting methods, a non - stationary time series prediction method based on wavelet transform and ARIMA - SVM Combined Model was proposed. By wavelet decomposition and reconstruction, the non - stationa- ry time se - ties were decomposed into a low frequency signal and several high frequency signals. The high frequency signals were predicted with auto - regression models ARIMA, and the low frequency was predicted with SVM. The prediction result of the original time series was the superimposition of the respective pre - diction. The method in this paper is better than the traditional rnelhods, and the obtained result in this paper is close to the actual value.

关 键 词：小波变换 非平稳时间序列 支持向量机组合模型 预测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]