计算机应用 ›› 2011, Vol. 31 ›› Issue (05): 1374-1377.DOI: 10.3724/SP.J.1087.2011.01374

• 数据库技术 • 上一篇    下一篇

面向多维时间序列的过程决策树模型

刘栋1,宋国杰2   

  1. 1.河南师范大学 计算机与信息技术学院,河南 新乡 453007
    2.北京大学 信息科学技术学院,北京 100871
  • 收稿日期:2010-10-26 修回日期:2010-12-15 发布日期:2011-05-01 出版日期:2011-05-01
  • 通讯作者: 刘栋
  • 作者简介:刘栋(1976-),男,河南原阳人,讲师,博士研究生,主要研究方向:决策支持系统、机器学习;宋国杰(1975-),男,河南原阳人,副教授,博士,主要研究方向:数据挖掘、机器学习。
  • 基金资助:

    国家自然科学基金资助项目(60703066)。

Process decision tree model based on multi-dimensional time series

LIU Dong1, SONG Guo-jie2   

  1. 1.College of Computer and Information Technology, Henan Normal University, Xinxiang Henan 453007, China
    2.College of Electronics Engineering and Computer Science, Peking University, Beijing 100871, China
  • Received:2010-10-26 Revised:2010-12-15 Online:2011-05-01 Published:2011-05-01
  • Contact: Liu Dong

摘要: 为解决多维时间序列的分类并获取易于理解的分类规则,引入了时序熵的概念及构造时序熵的方法,基于属性选择和属性值划分两方面扩展了决策树模型。并给出了两种构造多维时间序列分类的决策树模型算法。最后,采用移动客户流失的真实数据,对过程决策树进行测试,展示了方法的可行性。

关键词: 多维时间序列分类, 熵, 决策树, 分类规则

Abstract: To solve the classification problem of multi-dimensional time series and obtain understandable classification rules, the concept of time series entropy and the method of structuring time series entropy were introduced. And the decision tree model was expanded based on both attribute selection and attribute value. Two algorithms for structuring decision tree model of multi-dimensional time series classification were presented. Finally, process decision tree was tested on mobile customer churn data, and the feasibility of the proposed method was demonstrated.

Key words: multi-dimensional time series classification, entropy, decision tree, classification rule