SVR算法在指数预测中的应用研究

Study on the Application of SVR's Algorithm in Index Forecasting

本文阐述了回归型支持向量机(SVR)的基本结构及训练方法,并在此基础上研究了基于SVR算法的股票指数预测方法。通过应用LS-SVM软件,选用RBF核函数,利用自学模型,对超参数不断进行优化,以加快运算速度,并最终建立了该算法应用于股市预测的模型。通过股票指数的建模与仿真结果表明,支持向量回归机在股票价格的中短期预测以及整体股票趋势预测有比较好的效果。

This paper describes the basic structure and the training methods of Support Vector Regression （SVR）, on the basis of which, the Stock Index forecasting methods based on the SVR algorithm was researched. Through the application of IS- SVM, we study on the method of optimizing the super - Parameters with the RBF kernel constantly to speed up the computing speed, and then employ the algorithm to establish a stock market forecast model. The Modeling and the simulation result of Stock Index shows that Support Vector Regression has better effects in the medium- short term forecasting of the stock price as well as the forecasting of the whole tendency of stocks.