哈尔滨商业大学学报（自然科学版）2023，Vol.39Issue(6)：752-759,8.
基于DMD-Xgboost电商概念股的交易量化预测算法
Transaction quantitative prediction algorithm based on DMD-Xgboost e-commerce concept stocks
摘要
Abstract
Traditional machine learning methods lack the research of intrinsic associations of feature indicators based on market and high-dimensional nonlinear data characteristics,and only combine the data set with subjective and objective factors generally.Furthermore,previous research mostly concentrates on a single stock and the application scope was limited,so the quantitative trading prediction model based on DMD-Xgboost e-commerce concept stocks was proposed.The data set used 12 financial indicators of daily trading data of e-commerce concept stocks,eliminating the interference of human factors in quantitative trading by the DMD algorithm.Using the Ritz eigenvalue to find the most influential indicators as the input of the Xgboost model,which solve the insensitivity of machine learning to non-linear financial market information.The empirical results showed that this algorithm could effectively identify stock trading information with an RMSE value of about 0.117 8.The evaluation system was established and compared with the traditional model,indicating that the algorithm achieves an obvious improvement in prediction effect.关键词
Xgboost/DMD/电商概念股/学科交融/机器学习/股票市场Key words
Xgboost/DMD/e-commerce concept stocks/discipline cross-fertilization/machine learning/stock market分类
管理科学引用本文复制引用
童珺仪..基于DMD-Xgboost电商概念股的交易量化预测算法[J].哈尔滨商业大学学报（自然科学版）,2023,39(6):752-759,8.基金项目
国家自然科学基金(No.11871115) （No.11871115）
