基于相空间重构的LS-SVM股票价格预测

Stock price forecast based on LS-SVM phase space reconstruction

提出一种基于相空间重构的最小二乘支持向量机(LS-SVM)的股票价格预测方法。采用混沌时间序列对股票价格数据进行相空间重构,应用贝叶斯框架对最小二乘支持向量机的参数选优。预测结果表明,该模型具有误差小、拟合程度高等优点,可适用于股票价格预测。

A phase space reconstructed forecasting method of stock price was proposed based on least squares support vector machines（LS-SVM）.The data of stock price was phase space reconstructed based on chaotic time series.The model of parameters was optimized by Bayesian framework.The forecasting results show that the method has the advantages of smaller error and higher fitting and is applicable in stock price forecast.