概念学习中连续值型属性的离散化

DISCRETIZING CONTINUOUS VALUED ATTRIBUTES IN CONCEPT LEARNING

在概念学习中，属性不仅可以为离散值型，还可以为连续值型，因此，连续值型属性的离散化问题是概念学习中的一个重要问题，本文给出了基于假设检验的离散化方法的理论依据，并依此提出了一种离散化算法Ｄｉｓｃｒｅｔｅ，实验结果表明，这种方法一般能得到较合理的区间划分．

Attributes can take not only discrete values but also continous values in concept learning. This article mainly discusses how to discretize continous valued attributes, and propose a discretizing method based on χ 2 test. The experimental result shows that this approach can get reasonable intervals.