一种基于主成分LSTM模型在股票预测中的研究  

A Study of Stock Forecasting Based on LSTM Model of Principal Component Analysis

作　　者：王东[1] 王霄鹏 杨川东 WANG Dong;WANG Xiaopeng;YANG Chuandong(College of Computer Science and Engineering,Chongqing University of Technology,Chongqing 400054,China)

机构地区：[1]重庆理工大学计算机科学与工程学院,重庆400054

出　　处：《重庆理工大学学报（自然科学）》2021年第2期282-288,共7页Journal of Chongqing University of Technology：Natural Science

基　　金：重庆市教委雏鹰计划研究项目(CY180903)。

摘　　要：在利用技术方法建立LSTM股票预测模型时,传统方法由于所选择的输入数据变量较多、数据信息存在重叠、异常值对训练影响较大等因素,经常导致泛化性差,预测效果欠佳。针对此类问题,提出利用主成分分析法将基础数据降维,再结合股票相关技术指标KDJ,MACD一同作为输入数据,并根据股票特性将模型调整后再进行预测。实验结果表明:PCA-S-LSTM模型在降低预测平均误差的同时,大大减少了运行时间,提高了预测稳定性,较为准确地预测了平安银行的收盘价,具有应用价值。Stock is well-known as a multi-element complex system.The input vectors have a large impact on training speed and prediction results when using a technical method to establish the LSTM forecast model of stock.Due to many selected input vectors,overlaps between data information,great influence of outliers to training and other factors,the prediction effect in traditional methods is usually not ideal.In order to solve such problems,this paper proposes to use principal component analysis to reduce the dimension of basic data,combine with stock-related technical indicators KDJ and MACD as input data,and adjust the model according to the characteristics of the stock before forecasting.The experimental results show that the PCA-S-LSTM model reduces the average error of prediction,meanwhile reduces the running time greatly,improves the stability of prediction,and accurately forecasts the closing price of Ping An Bank.Therefore,the PCA-S-LSTM model has been proved to have a certain application value.

关 键 词：神经网络 主成分分析法 LSTM模型 股票价格预测 

分 类 号：TP138[自动化与计算机技术—控制理论与控制工程]