基于Voronoi和空间自相关的离群点检测

Outlier Detection Based on Voronoi and Spatial Autocorrelation

为了提高空间数据挖掘的效率和准确度,在分析传统的离群点检测算法优、缺点的基础上,提出一种空间离群点检测算法。用Voronoi来确定空间对象间的邻近关系,在空间邻域内利用空间自相关性来计算局部Moran指数,并将其作为离群因子进而判断离群点。实验结果表明,该算法能够高效、准确地检测出空间离群点,具有对用户依赖性少和可伸缩性强等优点。

In order to improve the spatial data mining efficiency and accuracy, the research on spatial outlier detection algorithm based on Voronoi and spatial autocorrelation is proposed after analyzing the advantages and disadvantages of the classical outlier detection algorithms. The algorithm calculates local Moran index of non-spatial attribute as the outlier factor by Voronoi neighborhoods without parameter. Experimental results show that the proposed algorithm can outperform other existing algorithms in detection accuracy, user dependency and efficiency.