摘要
采用贡献度与相关性分析(correlation analysis)相结合的办法从目前最常用的244种股票技术指标中提取最优技术指标,进而利用梯度提升树(GBDT)算法对股票的趋势进行预测.对由贡献度、相关性分析与GBDT算法构成的组合模型(简称GBDT组合模型)进行实证分析,首次将GBDT算法应用于沪深300股票的预测.对由不同算法构成的组合模型的预测精度也进行比较分析.实验结果表明,GBDT组合模型在预测精度上优于线性回归组合模型及随机森林组合模型.
The article uses the combination of contribution degree and correlation analysis to extract the optimal technical in-dex from the 244 most commonly used stock technical indices,and then uses gradient boosting decision tree(GBDT)to forecast the trend of the stock.Empirical analysis of the combination model consisting of the contribution degree,correla-tion analysis and GBDT algorithm(referred to as GBDT combination model)is made,and for the first time the GBDT algo-rithm is applied to predict the CSI 300 stock set.The prediction accuracy of the combined model composed of different algo-rithms is also compared and analyzed.The experimental results show that the GBDT combination model is superior to the linear regression combination model and random forest combination model in prediction accuracy.
作者
张潇
韦增欣
杨天山
ZHANG Xiao;WEI Zengxin;YANG Tianshan(School of Mathematics and Information Science,Guangxi University,Nanning 530004,China;School of Finance and Economics,Nanning College for Vocational Technology,Nanning 530004,China)
出处
《海南师范大学学报（自然科学版）》
CAS
2018年第1期73-80,共8页
Journal of Hainan Normal University（Natural Science）
基金
国家自然科学基金(11161003)
中青年教师基础能力提升项目(2017KY1017)
关键词
GBDT
股票预测
相关性分析
贡献度
股票技术指标
GBDT
stock forecast
correlation analysis
contribution degree
stock technical index
