|
|||||
|
|
| 基于相似系数和的孤立点检测算法 | |
| 引用本文： | 陆声链,林士敏,黄万华. 基于相似系数和的孤立点检测算法[J]. 安徽师范大学学报(自然科学版), 2004, 27(2): 133-136 |
| 作者姓名： | 陆声链 林士敏 黄万华 |
| 作者单位： | 1. 广西师范大学,数学与计算机科学学院 2. 广西师范大学,教务处,广西,桂林,541004 |
| 摘 要： | 简要介绍了一个基于相似系数和的孤立点检测聚类算法,指出了它的局限性,在此的基础上,文中提出了一个改进的算法.算法的思想是,对数据集进行标准化,然后构造一个相似系数矩阵,通过对象的相似系数之和判断对象的孤立程度.改进后的算法除了可以检测出倍数异常孤立点外,还可以检测出分量异常孤立点. |
| 关 键 词： | 字:孤立点检测 孤立点 相似数 |
| 文章编号： | 1001-2443(2004)02-0133-04 |
| 修稿时间： | 2003-10-28 |
AN ALGORITHM FOR OUTLIER DETECTION BASED ON SIMILAR COEFFICIENT SUM |
|
| LU Sheng-lian, LIN Shi-min, HUANG Wan-hua. AN ALGORITHM FOR OUTLIER DETECTION BASED ON SIMILAR COEFFICIENT SUM[J]. Journal of Anhui Normal University(Natural Science Edition), 2004, 27(2): 133-136 | |
| Authors: | LU Sheng-lian LIN Shi-min HUANG Wan-hua |
| Affiliation: | LU Sheng-lian~1, LIN Shi-min~1, HUANG Wan-hua~2 |
| Abstract: | After introducing simply a clustering algorithm for checking outlier based on similar coefficient sum, the paper points out its defect. Basing on this, the paper proposes a improved algorithm. Its ideas can be listed as follows: normalizing the attributes data set, then creating a similar coefficient matrix, the degree of deviation can be judged by object's similar coefficient sum. The improved algorithm not noly can detects outliers of times anomaly, but also detects outliers of component anomaly. |
| Keywords: | outlier detection outlier similar coefficient |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ | |