基于情感分析和Informer的股票预测方法  

Stock Prediction Method Based on Sentiment Analysis and Informer

作　　者：倪学曜 NI Xueyao(School of Information Engineering,Huzhou University,Huzhou 313000,China)

机构地区：[1]湖州师范学院信息工程学院,浙江湖州313000

出　　处：《现代信息科技》2025年第4期139-144,共6页Modern Information Technology

摘　　要：针对股票价格受复杂因素及投资者情绪影响导致的预测难题,提出了一种融合情感分析与Informer模型的股票价格预测方法。首先,运用AdaBoost模型从大量数据中提取关键特征变量,以降低模型过拟合的风险。其次,构建金融情感词典,对通过网络爬虫获取的金融文本数据进行情感分析,并计算情绪指数。最后,利用Informer模型对中信证券、华泰证券、东方财富三只具有代表性的证券龙头股票进行预测。评估结果显示,该方法有效提升了预测的准确性,验证了其有效性和实用性。Aiming at the prediction problem of stock price caused by complex factors and investor sentiment,a stock price prediction method based on sentiment analysis and Informer model is proposed.Firstly,the AdaBoost model is used to extract key feature variables from a large amount of data to reduce the risk of model overfitting.Secondly,a financial sentiment dictionary is constructed to analyze the sentiment of the financial text data obtained by the web crawler,and the sentiment index is calculated.Finally,the Informer model is used to predict three representative security leading stocks of CITIC Securities,Huatai Securities and Eastmoney Securities.The evaluation results show that the method effectively improves the accuracy of prediction,and verifies its effectiveness and practicability.

关 键 词：情感分析 股票预测 Informer模型 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TP391[自动化与计算机技术—控制科学与工程] F832.5[经济管理—金融学]