基于多特征融合的股票走势预测研究  

Research on stock trend prediction based on multi-feature fusion

作　　者：刘月娟 王武 LIU Yue-juan;WANG Wu(School of Mathematics and Computer Science,Yunnan Minzu University,Kunming 650500,China)

机构地区：[1]云南民族大学数学与计算机科学学院,云南昆明650500

出　　处：《云南民族大学学报（自然科学版）》2022年第2期227-234,共8页Journal of Yunnan Minzu University:Natural Sciences Edition

基　　金：国家自然科学基金(61962033);云南民族大学研究生创新基金(SJXY-2020-105).

摘　　要：为了更好地对股票走势进行预测,提出了一种在原有BiLSTM基础上引入注意力机制的股票预测模型,该模型不仅将股票交易数据作为模型输入,同时考虑到金融新闻对股价的影响.针对Reddit中苹果公司2010年到2018年间每日的新闻标题进行了研究,模型的输入特征来自于3部分:一是从文本数据中提取出的语义特征,二是股票的历史交易数据,三是从文本数据中提取出的情感特征,最后将上述输入特征融合到一起放入BiLSTM+Attention模型中进行训练,达到对股票的次日收盘价预测的效果.最终对5个模型进行对比实验,评价指标结果表明,提出的模型较其他模型相比预测效果更好.In order to predict the trend of stock price accurately,a novel prediction model which introduces an attention mechanism based on the original BiLSTM is proposed in this paper.We not only take stock transaction data as the input,but also take into account the impact of financial news on stock prices.In turn,we use daily news headlines of Apple from 2010 to 2018 in Reddit as the training data.The feature data which is put into the proposed BiLSTM+attentionmodell comes from three parts:one is semantic feature extracted from text data,the other is historical stock trading data,and the third is from the emotional feature extracted from the text data.Finally,a comparative experiment was carried out on the five models,and the evaluation index results show that the proposed model in this paper has better predictive effect than other models.

关 键 词：股票预测 双向长短期记忆网络(BiLSTM) 注意力机制 情感分析 

分 类 号：TP110.67[自动化与计算机技术—控制理论与控制工程]