摘要
针对C4.5决策树算法在处理多维数据分类时,没有考虑各属性对分类结果的影响,导致分类准确率低的问题,提出一种基于距离权值的C4.5组合决策树算法。根据标准欧式距离定义数据属性的距离权值,更新C4.5决策树算法的信息增益率,得到基于距离权值的C4.5算法。利用改进后的C4.5决策树分类算法训练多个基分类器,基分类器通过Bagging集成方法构建组合决策树。实验结果表明,该算法在处理多维数据时有较高的准确性和稳定性。
To address the problems of C4.5 decision tree algorithm ignoring the influences of different attributes on the classification results and having low accuracy on dealing with multidimensional data sets,the multiple classifiers of C4.5 decision tree algorithm based on the distance weight was proposed.By adopting the distance weight which was defined according to the stan-dard Euclidean distances,the information gain ratio was updated.Hence the distance weight based algorithm was presented.Elementary classifiers which were trained through the Bagging algorithm built new multiple classifiers.Results of simulation show that the proposed algorithm has higher accuracy and stability in dealing with multidimensional data sets.
出处
《计算机工程与设计》
北大核心
2018年第1期96-102,共7页
Computer Engineering and Design
基金
国家自然科学基金项目(41575155)
关键词
C4.5算法
标准欧式距离
距离权值
信息增益率
Bagging组合算法
C4.5 decision tree algorithm
standard Euclidean distance
distance weight
information gain ratio
Bagging multiple classifiers
