基于空间局部偏离因子的离群点检测算法

Outlier Detection Algorithm Based on Space Local Deviation Factor

针对空间数据集的特性,提出一种基于空间局部偏离因子(SLDF)的离群点检测算法。利用SLDF度量空间点对象的离群程度,计算空间数据集中点对象的SLDF值并对其进行排序,将取值较大的前M个点对象作为空间离群点。实验结果表明,该算法能较好地检测空间局部离群点,其有效性与准确性均优于SLZ算法,适用于高维大数据集的空间离群点检测。

According to the characteristics of spatial data sets, this paper proposes an outlier detection algorithm based on the Space Local Deviation Factor（SLDF）. The algorithm uses SLDF to measure the deviate degree of space points object. It calculates all the points＇ SLDE sorts by their values, and uses the top M as the space outlier. Experimental result shows that the algorithm can well detect space outlier and be more applicable to the high dimensional and large data sets, its validity and accuracy of the algorithm are superior to that of SLZ algorithm.