决策树剪枝加强的关联规则分类方法

传统关联规则挖掘在面临分类决策问题时，易出现非频繁规则遗漏、预测精度不高的问题。为得到正确合理且更为完整的规则，提出了一种改进方法 DT-AR(decision tree-association rule algorithm)，利用决策树剪枝策略对关联规则集进行补充。该方法利用FP-Growth(frequent pattern growth)算法得到关联规则集，利用C4.5算法构建后剪枝决策树并提取分类规则，在进行置信度迭代筛选后与关联规则集取并集修正，利用置信度作为权重系数采取投票法进行分类。实验结果表明，与传统关联规则挖掘和决策树剪枝方法相比，该方法得到的规则在数据集分类结果上更准确。

Association Rule Classification Method Strengthened by Decision Tree Pruning

Solving the problem of mining rules assisting decision making with traditional association rule comes along with omission of infrequent rules and the low prediction accuracy usuallily. In order to obtain more correct, reasonable and complete rules, a method, DT-AR（decision tree-association rule algorithm） is proposed, which supplements the association rule set by decision tree pruning strategy. Firstly, the method gets the association rule set by FP-Growth（frequent pattern growth）. And then it constructs the post pruning decision tree and extracts classification rules. After iterative confidence filtering, it makes corrections to the association rule set to take the union with the classification rule set. Finally, it makes data classification by voting method, with the conifidence as the weight. The experimental results show that the rules of DT-AR get more accurate results in the prediction compared with the method of association rule and decision tree pruning.