空间离群点的检测算法

A Spatial Outlier Detection Algorithm

空间离群点是指与其邻居具有明显区别的属性值的空间对象。已有的空间离散点检测算法一个主要的缺陷就是这些方法导致一些真正的离群点被忽略而把一些非离群点当成了空间离群点。提出了一种迭代算法,该算法通过多次迭代检测离群点,取得较好效果。实验表明该算法具有较好的实用性。

A spatial outlier is a spatially referenced object whose non - spatial attribute values are significantly different from the values of its neighborhood The identification of outliers can lead to the discovery of useful knowledge and detecting spatial oudiers is useful in many applications of geographic information systems. One major drawback of the existing detection approaches is that their application will lead ro some true spatial oudiers being ignored and some false spatial outliers being identified. In this paper, propose an iterative algorithm that detects spatial outliers by multi - iterations. This algorithm gets a good performance. Experiments show that this algorithm has a good practicality.