Associative Classifier for Uncertain Data

摘要：

Associative classifiers are relatively easy for people to understand and often outperform decision tree learners on many classification problems. Ex isting associative classifiers only work with certain data. However, data uncer tainty is prevalent in many real-world applications such as sensor network, market analysis and medical diagnosis. And uncertainty may render many conventional classifiers inapplicable to uncertain classification tasks. In this paper, based on U-Apriori algorothm and CBA algorithm, we propose an associative classifier for uncertain data, uCBA (uncertain Classification Based on Associative), which can classify both certain and uncertain data. The algorithm redefines the sup port, confidence, rule pruning and classification strategy of CBA. Experimental results on 21 datasets from UCI Repository demonstrate that the proposed algo rithm yields good performance and has satisfactory performance even on highly uncertain data.

关键词： Associative Classification Uncertain Data Multiple Rules Classification Expected Support

作者: Xiangju Qin Yang Zhang Xue Li Yong Wang

作者单位: College of Information Engineering, Northwest A&F University, ER. China School of Information Technology and Electrical Engineering, The University of Queensland, Australia School of Computer, Northwest Polytechnical University, P.R. China

会议类型: 国际会议

会议名称: 11th International Conference,WAIM 2010(第十一届网络时代管理国际会议)

会议地点: 九寨沟

会议语种:英文

页码: 692-703

在线出版日期: 2010-07-14（万方平台首次上网日期，不代表论文的发表时间）