计算机应用研究2016，Vol.33Issue(12)：3700-3704,5.DOI:10.3969/j.issn.1001-3695.2016.12.040
基于地统计学的空间离群点检测算法的研究
Geographical statistics based approach for spatial outlier detection
摘要
Abstract
In view of the traditional spatial outliers detection algorithms have difficult in parameter selection when constructing neighborhood and have high time complexity when dealing high-dimensional data,this paper proposed a spatial outliers detec-tion algorithm based on the geographical statistical theory.This algorithm introduced the spatial autocorrelation theory into the spatial outlier detection,firstly used 3σrules to identify global outliers,and then used Delaunay triangulation neighborhood rela-tionship building spatial neighborhood relationship,instead the global outliers by the average neighborhood node,finally used local Moran’I as space abnormal measure method.The simulation results show that the method does not have to select parame-ters,strong robustness,higher detection rate and lower false alarm rate.关键词
地统计学/空间离群点/Delaunay三角网/局部Moran’IKey words
geographical statistics/spatial outliers/Delaunay triangulation/local Moran’I分类
信息技术与安全科学引用本文复制引用
刘莘,张绍良,王飞,张赛男..基于地统计学的空间离群点检测算法的研究[J].计算机应用研究,2016,33(12):3700-3704,5.基金项目
国家“十二五”科技支撑计划资助项目（2011BAC08B03）；江苏省煤基CO2捕集与地质储存重点实验室（中国矿业大学）开放基金资助项目（2015A01）；江苏高校优势学科建设工程资助项目 （）
