基于EEMD-Elman-Adaboost的中美股票价格预测研究  

Research on Stock Price Prediction of China and America Basedon EEMD-Elman-Adaboost

作　　者：杨静凌 唐国强[1] 张建文 YANG Jing-ling;TANG Guo-qiang;ZHANG Jian-wen(College of Science,GuiLinuniversity of technology,GuiLin 541006,China)

机构地区：[1]桂林理工大学理学院,广西桂林541006

出　　处：《运筹与管理》2022年第11期194-199,共6页Operations Research and Management Science

基　　金：国家自然科学基金资助项目(71963008)。

摘　　要：针对股票价格序列高度非正态、非线性、非平稳等复杂特征,文章以Elman神经网络为基础,引入集合经验模态分解(EEMD)与Adaboost算法,对中美股票的日收盘价进行预测。首先,利用EEMD算法将样本分解为多个本征模函数分量和1个残差分量。其次,用Adaboost算法优化Elman神经网络,对各个分量进行预测。最后,将各分量预测结果进行求和,作为最终预测结果。研究结果表明:EEMD-Elman-Adaboost模型对中美股票价格预测的均方根误差、平均相对误差、平均绝对误差均比现有的BP、Elman、EMD-Elman、EEMD-Elman模型小,新组合模型融合了EEMD、Elman神经网络、Adaboost算法的优点,具有更强的泛化能力和跟随能力。Aiming at the complex features such as highly non-normal,non-linear,non-stationary characteristics of stock price series,this paper introduces ensemble empirical mode decomposition(EEMD)and Adaboost algorithm based on Elman neural network to predict the daily closing price of Chinese and American stocks.Firstly,the EEMD algorithm is used to decompose the sample data into several intrinsic mode function(IMF)components and a residual component.Secondly,the Elman neural network is optimized with the Adaboost algorithm to make rolling predictions for each component.Finally,the sum of the prediction results of each component is used as the final prediction result.The results show that the root mean square error,mean absolute percentage error,and mean absolute error of EEMD-Elman-Adaboost model for predicting the stock prices of China and America are smaller than those of existing BP,Elman,EMD-Elman and EEMD-Elman models.The new combination model integrates the advantages of EEMD,Elman neural network and Adaboost algorithm,so that it has stronger generalization ability and following ability.

关 键 词：股票收盘价 EEMD ELMAN ADABOOST 组合模型预测 

分 类 号：F224[经济管理—国民经济]