一种多变量决策树方法研究  

A New Multivariate Decision Tree Algorithm

作　　者：梁道雷[1] 黄国兴[1] 金健[1] 

机构地区：[1]华东师范大学计算机系

出　　处：《计算机科学》2008年第1期211-212,共2页Computer Science

基　　金：华东师范大学211重点项目(521B0108)

摘　　要：单变量的决策树算法造成树的规模庞大,规则复杂,不易理解。本文结合粗糙集原理中的相对核及加权粗糙度的方法,提出了一种新的多变量决策树算法。通过实例表明,本文的多变量决策树方法产生的决策树比传统的ID3算法构造的决策树更简单,具有较好的分类效果。Decision Tree Algorithm in univariate tests caused large-scale, complex rules that are difficult to understand. Based on the rough sets theory of attributes reduction, the core of condition attributions and the Weighted roughness Of condition attributions, a new multivariate decision tree algorithm is proposed. A example shows in this paper, the decision tree built by the method is more simple and has better classification result than that of ID3 algorithm.

关 键 词：多变量决策树 粗糙集 相对核 加权粗糙度 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]