摘要
在金融投资领域中有不同的选择,普通投资可以选择ETF基金进行资产配置。本文针对ETF基金时间序列数据,利用LSTM构建单变量预测模型,选择收盘价作为因子。实验数据选择了具有代表性的创中盘88ETF、军工ETF和消费ETF,对其进行了数据预处理。设置对照实验,对比卷积神经网络、时域卷积网络模型的预测结果。通过实验评价指标结果发现,LSMT相比对照实验模型,拥有更高的预测精确度。
In financial investment,there are different options,and ordinary investors can choose ETF funds for asset allocation.In this paper,we use the LSTM to construct a single variable prediction model for ETF fund time series data,and choose the closing price as a factor.The experimental data are selected from representative Chuanzhong Pan88ETF,military ETF,and consumer ETF,and data preprocessing is performed on them.The prediction experimental results are compared with the convolutional neural network and time domain convolutional network models.It can observe that LSMT has higher prediction accuracy than other methods.
作者
秦佳兵
芦立华
姬乘风
QIN Jiabing;LU Lihua;JI Chengfeng(School of Electric Information Engineering,Shanghai Dianji University,Shanghai,China,201306)
出处
《福建电脑》
2023年第8期15-19,共5页
Journal of Fujian Computer