基于EWT-PSO-SVM误差校正组合模型的中国股市预测研究

崔金鑫,邹辉文

系统科学与数学 ›› 2019, Vol. 39 ›› Issue (8) : 1212-1235.

PDF(3347 KB)
PDF(3347 KB)
系统科学与数学 ›› 2019, Vol. 39 ›› Issue (8) : 1212-1235. DOI: 10.12341/jssms13689
论文

基于EWT-PSO-SVM误差校正组合模型的中国股市预测研究

    崔金鑫1,2,邹辉文1,2
作者信息 +

Prediction of China Stock Market Based on EWT-PSO-SVM Error Correction Combination Model

    CUI Jinxin 1,2 ,ZOU Huiwen 1,2
Author information +
文章历史 +

摘要

鉴于股市预测的复杂性, 遵循``先分解后集成''的总体建模思路, 文章基 于EWT分解算法和SVM支持向量机模型, 同时结合PSO粒子群优化算法和误差校正组 合预测方法, 构建了一种中国股票市场建模及预测的EWT-PSO-SVM误差校正组合预 测模型. 先基于EWT算法将原始价格序列分解成若干分量, 再根据频率将其重组成高、中、低频3个分量, 对它们分别建立PSO-SVM误差校正组合模型, 最后集成各个分量 的预测结果. 与其他预测模型相比较, 文章所构建预测模型的MSE、MAE、MAPE、 RMSE、Theil不等系数、确定性系数DC和方向性指标DS 7个指标均优于其他基准预 测模型, MCS检验结果同样表明本文构建模型的预测性能最优, 稳健性检验结果进一步证实了文章构建的模型预测性能所具备的稳健性.

Abstract

Given the complexity of stock market prediction, following the modeling approach of ``decomposition first and integration last'', based on EWT decomposition algorithm and SVM support vector machine model, combined with PSO particle swarm optimization algorithm and error correction combination forecasting method, an EWT-PSO-SVM error correction combined forecasting model for Chinese stock market modeling and prediction is proposed. Firstly, the original price series is decomposed into several components based on EWT algorithm, and then they are reconstructed into high, medium and low frequency components according to the frequency. The error correction combination model is established respectively, and the prediction result is finally obtained. Compared with other prediction models, the MSE, MAE, MAPE, RMSE, Theil inequality coefficient TIC, deterministic coefficient DC and direction indicator DS of the model proposed in this paper are superior to other benchmark forecasting models, the MCS test results also support this conclusion. The robustness test results further reflect the robustness of the proposed model.

关键词

EWT算法 / PSO算法 / SVM模型 / 误差校正组合模型 / 股市预测.

引用本文

导出引用
崔金鑫 , 邹辉文. 基于EWT-PSO-SVM误差校正组合模型的中国股市预测研究. 系统科学与数学, 2019, 39(8): 1212-1235. https://doi.org/10.12341/jssms13689
CUI Jinxin , ZOU Huiwen. Prediction of China Stock Market Based on EWT-PSO-SVM Error Correction Combination Model. Journal of Systems Science and Mathematical Sciences, 2019, 39(8): 1212-1235 https://doi.org/10.12341/jssms13689
PDF(3347 KB)

Accesses

Citation

Detail

段落导航
相关文章

/