粗糙集在决策树生成中的应用

决策树是归纳学习的重要形式，建造高质量的决策树的关键是选择合适的属性，本文针对ＩＤ３算法对属性间的相依性强调不够等问题，利用粗糙集理论，提出了一种新的启发式函数－分辩率构造决策树。

Application of Rough Sets to the Designing of Decision Trees

Decision trees are important forms of inductive learning. The key to building a good decision tree lies in the reasonable choice of attributes. In relation to problems existing in ID3 algorithm, such as overlooking the interconnection between attributes, the paper proposes a new super attribute, resolution, for the optimization of decision trees based on rough sets. The super attribute, in essence, is the integration of relational attributes, but not simple integration. It stresses both the dependency of attributes and the number of classification. Some examples are given to show that the method herein is obviously better than ID3 algorithm.