基于遗传聚类算法的离群点检测  

Outlier detection based on genetic algorithm for clustering

作　　者：钱光超[1] 贾瑞玉[1] 张然[1] 李龙澍[1] 

机构地区：[1]安徽大学计算机科学与技术学院,合肥230039

出　　处：《计算机工程与应用》2008年第11期155-157,共3页Computer Engineering and Applications

基　　金：安徽省教育厅资助科研课题(the Research Project of Department of Education of Anhui Province; China under Grant No.2005KJ056)

摘　　要：离群点检测是数据挖掘一个重要内容,它为分析各种海量的、复杂的、含有噪声的数据提供了新的方法。对离群数据挖掘几类主要的方法进行了分析和评价,并在此基础上了提出了一种基于遗传聚类的离群点检测算法。该算法结合了遗传算法全局搜索的优点和K-均值方法局部收敛速度快的特点,取得较好效果。实验验证该算法很好地检测到数据集中的离群点,同时还完成了数据集的聚类。具有较好的实用性。Outlier detection, as an important aspect of data mining, provides a new method for analyzing various quantitative,complex and noisy data.In this paper,authors analyze and evaluate several major methods of the outlier data mining,and propose a new outlier detection algorithm which is based on an genetic algorithm for clustering.By integrating with global searching of the genetic algorithm and the good local convergence rate of the K-means algorithm,this algorithm gets a better result.Experiments show that this algorithm not only can detect the outliers in the dataset,but also complete the clustering of the dataset.So it has a good practicality.

关 键 词：离群点检测 数据挖掘 遗传算法 聚类 K-均值算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]