Abstract: This paper puts forward a new definition for the comprehensibility of a classification rule. The paper also gives a method for computing a rule’s comprehensibility by using attribute’s information gain. The method of mining classification rules using GA is also analyzed. The experiment on breast cancer statistical data shows that it can find accurate as well as comprehensible rules.

Key words: Classification rules, Genetic algorithm, Information gain, Comprehensible rules

摘要： 提出了一种分类规则易于理解性的新的定义，并给出了应用属性信息增益计算分类规则可理解性的方法。分析了遗传算法发现分类规则的过程。乳腺癌统计数据上的实验表明，它可以发现准确和易于理解的分类规则。

关键词: 分类规则, 遗传算法, 信息增益, 易于理解的规则