| 决策树算法的研究与改进 |
| |
| 引用本文： | 冯少荣.决策树算法的研究与改进[J].厦门大学学报(自然科学版),2007,46(4):496-500. |
| |
| 作者姓名： | 冯少荣 |
| |
| 作者单位： | 厦门大学计算机科学系 |
| |
| 基金项目： | 福建省自然科学基金;福建省高新技术项目 |
| |
| 摘 要： | 决策树是数据挖掘中重要的分类方法，本文在研究和比较几种经典的决策树算法基础上，提出了一种改进的决策树算法：基于度量的决策树（MBDT）．这种决策树实际上是把线性分类器和决策树结合在一起．实验证明，用该方法构造的决策树能有效地减少决策树的层数，从而提高决策树的分类效率．通过MBDT分类实验，验证了上面结论的正确性和有效性．
|
| 关 键 词： | 决策树 度量 ID3算法 熵 |
| 文章编号： | 0438-0479（2007）04-0496-05 |
| 修稿时间： | 2006-06-01 |
Research and Improvement of Decision Trees Algorithm |
| |
| FENG Shao-rong.Research and Improvement of Decision Trees Algorithm[J].Journal of Xiamen University(Natural Science),2007,46(4):496-500. |
| |
| Authors: | FENG Shao-rong |
| |
| Institution: | School of Computer Science and Engineering,South China University of Technology, Guangzhou 510641,China |
| |
| Abstract: | Decision tree is a key classification method in data mining. Firstly,based on the research and comparison of several classic decision trees algorithms, an improved decision tree algorithm based on metric is proposed in this paper. In practice, this kind of decision trees combines linear classifier and decision trees. The experimental result indicates that the decision trees based on this method can effectively reduce the decision trees level, which enhance classified efficiency of the decision trees. MBDT classified experiment results demonstrate the correctness and availability of the above conclusions. |
| |
| Keywords: | decision trees metric ID3 algorithm entropy |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
