文献类型：期刊文献

英文题名：Stock Indexes Forecasting Based on CNN-GRU-XGBOOST Hybrid Model

作者：Zhu, Wei[1]; Hu, QingChun[1]; Wang, MingCheng[1]; Hu, Kai[1]

机构：[1] East China University of Science and Technology, Shanghai, China

年份：2024

起止页码：195

外文期刊名：ACM International Conference Proceeding Series

收录：EI(收录号：20243616988128)

语种：英文

外文关键词：Financial markets

摘要：Stock indexes are key indicators of overall stock market. Accurate forecasting of stock indexes not only improves investment returns, but also relates to the development of financial markets. However, the high complexity of stock market challenges the prediction models. To get better prediction, hybrid model is a feasible scheme. The article proposes a hybrid model named CNN-GRU-XGBOOST for predicting stock indexes. Train the CNN-GRU and XGBOOST models with the training data firstly. Then predict the validation set by CNN-GRU and XGBOOST to acquire the forecasting outcomes of the two models. After that, treat these two prediction results as independent variable X and the label of the validation set as dependent variable Y. Subsequently, a linear regression is performed between X and Y. Finally, the linear regression parameters obtained in the previous step are used to fit the two test set results predicted by CNN-GRU and XGBOOST. In this way, the final prediction result is obtained. Through the prediction of Shanghai Composite Index, Shenzhen Component Index and Hong Kong's Hang Seng Index, the CNN-GRU-XGBOOST model performs better than the contrast model. ? 2024 ACM.