摘要： 针对离群点的挖掘，在现有的LOF 算法的基础上，提出了一种基于网格聚类技术的离群点挖掘算法AOMGC。该算法将离群点挖掘分成两步挖掘过程。此外，该算法对其网格的划分加以改进，并能根据数据信息自动生成划分间隔，从而提高了数据挖掘的效率。实验结果表明AOMGC 算法是可行的和有效的。

关键词: 数据挖掘；离群点；局部偏离因子；网格

Abstract: This paper aims at outlier mining, and proposes an algorithm of outlier mining called AOMGC based on grid clustering techniques, withthe existing algorithm of LOF. In this algorithm, the whole outlier mining is divided into two mining steps. In addition, this algorithm modifies themethods of grids partition cells. Also, it can automatically form partition intervals according to the data information, which enhances the efficiencyof data mining. The results of experiments indicate that AOMGC is adoptable and effective.

Key words: Data mining; Outliers; Local outlier factor; Grid