学术论文

      基于定量关联规则树的分类及回归预测算法

      Categorization and regression algorithm based on the quantitative association rule tree

      摘要:
      为了解决基于Apriori的分类关联规则算法挖掘数值型数据时效率和准确率偏低的问题,提出基于定量关联规则树的分类及回归预测算法。采用改进的定量关联规则算法挖掘数值型数据生成关联规则库,并基于关联规则树结构实现分类及回归预测。研究结果表明:改进的Apriori定量关联规则挖掘算法提高了分类预测的准确率并降低了计算复杂度;而采用关联规则树结构可使分类与回归预测时间明显加快,提高了样本匹配学习的速度。
      Abstract:
      To solve the problem of the low efficiency and accuracy of numerical data mining based on the Apriori categorization association rule algorithm, this article introduces a categorization and regression algorithm based on the quantitative association rule tree. The modified quantitative association rule algorithm is adopted to mine numerical datasets to generate an association rule base, and the association rule tree ( QART) is reconstructed to realize the categorization and regression prediction. The results show that quantitative association based on the modified Apriori algorithm is helpful for improving the accuracy of categorization and regression and reducing the computational complexity, and the quantitative association rule tree can improve the efficiency of categorization and regression and increase the rule matching speed.
      Author: WANG Ling LI Shu-lin WU Lu-lu
      作者单位: 北京科技大学自动化学院,北京100083; 北京科技大学钢铁流程先进控制教育部重点实验室,北京100083
      刊 名: 工程科学学报 ISTICEIPKU
      年,卷(期): 2016, 38(6)
      分类号: TP311
      机标分类号: TP3 TP1
      在线出版日期: 2016年10月14日
      基金项目: 国家自然科学基金资助项目,中央高校基本科研业务费资助项目,北京科技大学研究生教材专项基金