基于概率后缀树的股票时间序列预测方法研究  

Research of stock time series based on probabilistic suffix tree

作　　者：程小林[1] 郑兴[1] 李旭伟[1] 

机构地区：[1]四川大学计算机学院,成都610065

出　　处：《四川大学学报（自然科学版）》2018年第1期61-66,共6页Journal of Sichuan University(Natural Science Edition)

基　　金：国家自然科学基金(61173099)

摘　　要：在时间序列符号化基础上,本文引入概率后缀树PST模型,构建基于时间序列符号化和概率后缀树相结合的股票预测模型.本文选择在沪深300的10支股票数据上将预测模型与传统的马尔科夫模型MM和自回归移动平均模型ARMA进行对比,结果显示本文提出的股票预测模型优于MM模型和ARMA模型,验证了本文所提出的预测模型在投资收益上的有效性.A stock forecasting model was introduced in this paper, which was based on the combination of time series symbolization and Probabilistic Suffix Tree （PST）. In addition, the Markov Model （MM） and the Auto Regressive Moving Average Model （ARMA） was compared with the forecasting model of this paper. The stock of 10 CSI 300 indices was used as the experimental sample. The results show that the stock forecasting model proposed in this paper is better than the MM model and the ARMA model,

关 键 词：股票数据挖掘 时间序列符号化 高斯混合模型聚类 概率后缀树 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]