计算机工程与应用 ›› 2014, Vol. 50 ›› Issue (20): 139-143.

• 数据库、数据挖掘、机器学习 • 上一篇    下一篇

基于改进极限学习机算法的股票价格在线预测

陆  玉,张  华   

  1. 阜阳职业技术学院,安徽 阜阳 236031
  • 出版日期:2014-10-15 发布日期:2014-10-28

Online prediction of stock price based on improved extreme learning machine

LU Yu, ZHANG Hua   

  1. Fuyang Vocational and Technical College, Fuyang, Anhui 236031, China
  • Online:2014-10-15 Published:2014-10-28

摘要: 为了对股票价格进行准确、快速的在线预测,提出一种基于改进极限学习机算法(IELM)的股票价格在线预测模型。在极限学习机(ELM)中引入Cholesky分解方法,使网络权值随新样本的逐次加入递推更新,提高模型的泛化能力,加快网络学习效率,然后对交通银行股票(601328)的收盘价进行仿真实验。结果表明,相对于对比模型,IELM不仅提高了计算效率,而且提高了股票价格预测精度,可以实现股票价格快速、准确在线预测。

关键词: 股票价格, 极限学习机, 在线预测, 网络权值

Abstract: In order to conduct fast and accurate online prediction of stock price, a novel online prediction model of stock price based on Improved Extreme Learning Machine(IELM) algorithm is proposed. Cholesky factorization is introduced into Extreme Learning Machine(ELM) and weight can update based on sequential input new samples, and improve computational efficiency, and then simulation experiment is carried out on shares(601328). The results show that the proposed model can improve computational efficiency and improve prediction precision of stock price, so the proposed model can be used in stock price online prediction.

Key words: stock price, Extreme Learning Machine(ELM), online prediction, network weight