基于聚类ORh方法的异常值检测应用研究

Application of outlier detection based on ORh method

聚类方法尝试着通过相似于其它事例的事例形成聚类来寻找数据集观察值的正常分组.聚类和异常值检测之间有很强的关联性,特别是基于观察值之间距离的概念.基于聚类ORh方法应用于异常值检测中,其结果给出一组异常值概率排序.这个排序可以让公司以优化的方法应用于检查活动,并获得了很好的性能.

Clustering methods were tried out to find the natural groupxngs u forming clusters of cases that are similar to each other.There are strong relationships between the clustering and the outlier detection. The ORh method on outlier detection was adopted to provide a kind of fraud probability ranking as an outcome of the process. These rankings help the company to apply its inspection resources in an optimal way.