文献类型：期刊文献

英文题名：Design and Application of a Machine Learning Model for Quantitative Stock Selection Incorporating Association Analysis and Hierarchical Attention Mechanisms

作者：Zhang, Yuheng[1];Li, Xin'e[2]

第一作者：Zhang, Yuheng

通讯作者：Li, XE[1]

机构：[1]China Construct Bank, Private Banking Dept, Beijing, Peoples R China;[2]Beijing Union Univ, Management Coll, Beijing, Peoples R China

第一机构：China Construct Bank, Private Banking Dept, Beijing, Peoples R China

通讯机构：[1]corresponding author), Beijing Union Univ, Management Coll, Beijing, Peoples R China.|[1141755]北京联合大学管理学院;[11417]北京联合大学;

年份：2026

卷号：60

期号：2

起止页码：225-242

外文期刊名：ECONOMIC COMPUTATION AND ECONOMIC CYBERNETICS STUDIES AND RESEARCH

收录：;WOS:【SSCI(收录号:WOS:001811287500012)，SCI-EXPANDED(收录号:WOS:001811287500012)】；

语种：英文

外文关键词：association analysis; hierarchical attention mechanism; feature engineering; ensemble learning model; quantitative stock selection

摘要：The application of machine learning models to assist investors in quantitative stock selection is an emerging topic of interest. However, due to the complex and dynamic nature of stock markets and issues such as market inefficiency, single machine learning models may underperform, and existing ensemble methods risk losing useful information. To address these challenges, this paper designs and optimises a machine learning model tailored to the characteristics of the Chinese stock market and quantitative stock selection. The model incorporates association analysis into the feature engineering process, generating differentiated datasets through information addition. Furthermore, a hierarchical attention mechanism is introduced in the model integration phase, dynamically assigning weights to base models via "upward integration" and "downward allocation."The model's effectiveness is validated through comparative experiments, ablation studies, and hyperparameter analysis on a self-constructed dataset of over 3.34 million records. Backtesting further demonstrates its practical performance in the stock market, yielding promising results, as expected.