摘要
决策树算法的实现往往采用面向对象语言工具来实现,与数据库中的结构通常存在一定的差异,需要进行大量的数据转换。现在充分利用数据库中表结构特点和存储过程中PL/SQL语法的强大性及灵活性,采用一个动力计量计费系统中的数据,快速、有效且非递归地实现了决策树C4.5算法中的节点生成、扩展与剪枝主要过程;并进行了规则抽取。应用结果表明,该算法的实现方法具有一定的高效性、稳定性和普适性。
The oriented-object program-developing tools are usually used to implement the algorithms of decision tree. Since the data structures in the programming languages are different from these in a database,massive data conversion is needed. With the data from a power metrology-billing system,the critical steps of node generation,node extension and pruning in C4.5 algorithm of decision tree can be implemented quickly,efficiently and non-recursively by making full use of the features of table object and the flexibility of PL/SQL in stored procedure,and the corresponding rules can be abstracted easily. Experimental re-sults demonstrate that this method is effective,stable and adaptable.
出处
《现代电子技术》
2014年第8期91-94,98,共5页
Modern Electronics Technique
关键词
C4
5算法
信息增益
存储过程
动力计量计费系统
C4.5 Algorithm
information gain
stored procedure
power metrology-billing system
