北京航空航天大学学报 ›› 2007, Vol. 33 ›› Issue (03): 350-353.

• 论文 • 上一篇    下一篇

数字趋势序列的子序列匹配算法

贾素玲, 陈当阳   

  1. 北京航空航天大学 经济管理学院, 北京 100083
  • 收稿日期:2006-04-08 出版日期:2007-03-31 发布日期:2010-09-19
  • 作者简介:贾素玲(1954-),女,河南洛阳人,教授,jiasuling@buaa.edu.cn.
  • 基金资助:

    国家自然科学基金资助项目(70371007)

Subsequence matching algorithm between number trend sequences

Jia Suling, Chen Dangyang,   

  1. School of Economics and Management, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
  • Received:2006-04-08 Online:2007-03-31 Published:2010-09-19

摘要: 针对时序数据挖掘中传统趋势序列分析的缺点,提出了数字趋势序列、趋势序列展开等概念.根据数字趋势序列的特点,使用片段斜率所对应的弧度值来度量片段的趋势.针对数字趋势序列的子序列匹配问题,设计了DTW(Dynamic Time Warping)快速搜索算法.算法分为3个部分:DTW顺序搜索、约束机制、冗余消除机制.并使用实际的股票数据对算法进行了验证.

Abstract: To overcome the demerits of traditional trend sequences- analysis in temporal data mining, two concepts that were number trend sequences and trend sequences unwrapping were put forward. According to features of number trend sequences, radians were used to represent the trends of line segments. DTW-QS(dynamic time warping quick searching) algorithm was designed to solve the problem of subsequence matching between number trend sequences. The algorithm included three parts: DTW sequential searching, the mechanism of restriction and the mechanism of redundancy control, and DTW-QS algorithm was evaluated via experiments.

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发