基于有权重超图模型的离群点发现  

Research on Outlier Testing Method based on Weighted Hypergraph-model

作　　者：李威[1] 李宏岩[1] 米守防[1] 

机构地区：[1]大连民族学院计算机科学与工程学院,辽宁大连116605

出　　处：《大连民族学院学报》2006年第5期45-48,共4页Journal of Dalian Nationalities University

摘　　要：结合基于有权重支持度框架的关联规则挖掘方法和基于超图模型的离群点检测方法,给出了一种离群数据的改进定义,并通过一个简单的实例说明了这种离群数据的离群含义,且与原离群点定义做了比较,分析了新定义离群数据的应用价值.Outlier detection is a branch of Data Mining, and it has a wide application at present. This paper presents an approach to detect outlier. The approach is based on weighted association rule mining using weighted support and significant framework and multilevel hypergraph partitioning algorithms. This paper improves the definition of outlier of Finding Outliers in High - Dimensional Space. We test the approach in a real dataset and compare it with the original one.

关 键 词：数据挖掘 离群点 超图模型 权重 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]