改进的关联分类算法在信息化评估中的应用

在信息化评估过程中，传统关联分类算法无法优先发现短规则，且分类精度对规则次序的依赖较强。为此，提出基于子集支持度和多规则分类的关联分类算法，将训练集按待分类属性归类，利用子集支持度挖掘关联规则，通过计算类平均支持度对测试集进行分类。实验结果表明，该算法发现规则的能力和分类精度均优于传统方法。

Application of Improved Associative Classification Algorithm in Informatization Evaluation

In informatization evaluation,traditional Associative Classification（AC） algorithms can not give priority to find short rules and the classification accuracy highly depends on the sequence of the rules.To solve these problems,this paper proposes a new AC algorithm based on sub-support and multi-rules.The training set is classified according to the attribute of the class,and associative rules are found by comparing the sub-support,the test set is classified by calculating the average support of the rule set.Experimental result shows that the new algorithm has stronger ability on finding rules and higher classification accuracy than traditional methods.