| 一种改进的基于加权属性的SLOF离群点挖掘算法 |
| |
| 引用本文： | 赵向兵.一种改进的基于加权属性的SLOF离群点挖掘算法[J].雁北师范学院学报,2011,27(3). |
| |
| 作者姓名： | 赵向兵 |
| |
| 作者单位： | [1]太原科技大学计算机科学与技术学院,山西太原030024 ;[2]山西大同大学数学与计算机科学学院,山西大同037009 |
| |
| 摘 要： | SLOF算法采用了空间对象的空间属性和空间关系确定空间邻域,并结合非空间属性的权值来计算对象在其邻域内的离群度,但在计算属性权值时,仍然由邻域专家决定,存在人为因素。文中采用计算每个对象的每个非空间属性的去一划分信息熵增量,并通过这个值来反映各个属性对对象离群的贡献程度,给出一种改进的SLOF算法。实验结果表明,算法具有计算效率高和对用户依赖性小的优点。
|
| 关 键 词： | 高维数据 信息熵 息熵增量 属性权值 偏离因子 |
SLOF based on Weighted Attribute of the Improved Algorithm for Outlier Mining |
| |
| ZHAO Xiang-bing.SLOF based on Weighted Attribute of the Improved Algorithm for Outlier Mining[J].Journal of Yanbei Teachers College,2011,27(3). |
| |
| Authors: | ZHAO Xiang-bing |
| |
| Institution: | ZHAO Xiang-bing1,2(1.School of Computer Science and Technology,Taiyuan University of Science & Technology,Taiyuan Shanxi,030024,2.School of Mathematics and Computer Science,Shanxi Datong University,Datong Shanxi,037009) |
| |
| Abstract: | SLOF algorithm uses spatial properties and spatial relationships of spatial objects to determine spatial neighborhood and combined with the weight of non-spatial attributes to calculate the object＇s outlier in its neighborhood.However,the attribute weights are still determined by the neighborhood experts in the calculation,there are human factors.For each object,calculating leave-one division information entropy increment of each non-spatial attribute is introduced in this article when determining the right value of non-spatial attribute.This value can reflect the contribution of stray objects by the various attributes and furthermore an improved SLOF algorithm is given.Experimental results show that the algorithm has high computational efficiency and the advantage of less dependence on the user. |
| |
| Keywords: | high dimensional data entropy entropy increment attribute weights deviation factor |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
