连续属性离散化算法SHD及其改进

为了让规则抽取算法能更好地适用于连续属性领域的问题，文章提出了一种有导师的连续属性离散化算法ＳＨＤ，并将该算法扩展到多连续属性处理领域。在此基础上，文章对该算法的数据预处理过程进行了探讨，提出了一种基于类间离散度矩阵分析属性空间重构造算法，并将其应用到属性预处理过程中。算法测试证明，对于连续属性领域的问题，使用ＳＨＤ属性离散化算法将明显改进后继规则抽取算法的效果。

The Discretization of Continuous Features Algorithm SHD and its Improvement

In this paper,a new supervised discretization algorithm named SHD is designed that makes the successive rule-extraction more suited to the field of continuous features learning. Moreover,we improve it to process multi-continu- ous features,and we propose a feature space reconstruction algorithm,which is based on the analysis of inter-class dis- crete matrix,for feature preprocessing. The experiment shows the obviously improved the efficiency successive rule-ex- traction algorithm in the field of continuous features learning.