基于CNN及LSTM融合模型的上证指数预测
Prediction of Shanghai Stock Exchange Index Based on CNN and LSTM Fusion Model
李铖健 1孙海燕1
作者信息
- 1. 北京航空航天大学数学科学学院,北京 100191
- 折叠
摘要
在CNN以及LSTM的外接以及内嵌两种融合模型的基础上,依据上证指数特征对内嵌模型中的部分结构进行调整改进,并为步长参数选择提供充分的理论依据,同时综合考虑样本股特征,分别对上证指数及成份股数据构建融合预测模型,对上证指数的收盘价进行预测.多组模型的对比实验结果表明,所构建的融合模型能够更加准确地把握数据的结构特征与时序性质,自动挖掘数据内部的相关关系,实现上证指数中的准确预测,为金融研究中的模型选择提供一定参考.
Abstract
Based on the external and embedded fusion models of CNN and LSTM,some of the structures in the embedded models are adjusted and improved according to the characteristics of the SSE index,and provide a sufficient theoretical basis for the selection of step parameters.The comparison experimental results of multiple groups of models show that the constructed fusion model can grasp the structural characteristics and time-series nature of the data more accurately,and automatically explore the correlation relationships within the data to achieve accurate prediction in the SSE index,which provides some reference for the model selection in financial research.
关键词
深度学习/循环神经网络/卷积神经网络/融合模型/指数预测Key words
Deep learning/Recurrent neural network/Convolution neural network/Fusion model/Index forecast引用本文复制引用
出版年
2024
NETL
NSTL
万方数据