股票趋势预测分析研究——基于百度指数

Analysis of Stock trend forecasting--Based on Baidu Index

针对股票价格无规律、复杂的跌涨预测问题,考虑到投资者关注度可能对股票产生的影响,文章将百度旗下的百度指数作为投资者关注度的衡量标准,并确定与预测股票具有相关性的关键词。结合用于股票市场预测的神经网络模型——长短期记忆模型(LSTM模型),在对数据进行相关性分析、数据清理、数据归一化后,带入模型进行预测。实验结果表明:在不考虑宏观因素的情况下,找到有效的关键词作为投资者关注度的衡量指标,并带入模型中预测,不仅可以预测股票趋势,还能准确预测股票价格,让投资者在了解实际股价的情况下,作出适合的股票投资决策。

In view of the irregular and complex forecast of stock prices,considering the possible impact of investor’s attention on the stock,this paper takes the Baidu index of Baidu as the measurement standard of investor’s attention,and determines the keywords that are relevant to stock forecasting.Combined with the stock market prediction model of Neural Ensemble--short-term and long-term memory model(LSTM model),after data correlation analysis,data cleaning and data normalization,it was brought into the model for prediction.The experimental results show that without considering the macro factors,to find effective keywords as measuring indicators of investor’s attention and bring it into the model for prediction,can not only forecast the stock trend but also accurately predict the stock price,and let the investors make suitable decisions of stock investment when they know the actual stock price.