一种基于影响度的可伸缩的决策树算法  

A Retractile Decision Tree Algorithm Based on Influence Degree

作　　者：冯贵良[1] 杨洁[1] 李忠哗[1] 

机构地区：[1]河北北方学院理学院,河北张家口075000

出　　处：《河北北方学院学报（自然科学版）》2008年第4期55-57,61,共4页Journal of Hebei North University:Natural Science Edition

基　　金：河北省科技研究与发展指导项目(07213543)

摘　　要：目的针对ID3算法计算复杂度高这一问题,改进决策树生成算法DTA(Decision Tree Algorithm).方法提出了用影响度作为属性选择的标准,为了使算法具有良好的可伸缩性,引入了基于类别的属性表的新的数据结构.结果表明算法能生成正确的决策树,并且计算复杂度明显优于传统算法.结论可以在计算机硬件配置较低、资源消耗较少的条件下来快速生成正确的决策树,得到相应的决策规则.Objective Aiming at high complex degree of the algorithm ID3, to introduce an advanced algorithm, DTA （Decision Tree Algorithm）, which is based on decision tree. Methods Using influence degree as the standard of feature selection for a good calculation complex degree and using a new data structure which is an attribute list based on classification for a good retractility. Results As is shown from the result of experiment, the advanced algorithm can make accurate decision tree and better complex degree than traditional algorithms. Conclusion The algorithm can generate correct decision trees and obtain relevant decision rules on the conditions of low configuration of computer and less consumption of resourses.

关 键 词：决策树 数据挖掘 算法 属性表 影响度 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]