论文部分内容阅读
空间异常检测已成为空间数据挖掘和知识发现的一个重要研究内容.空间异常蕴含着许多意想不到的知识,现有的空间异常检测方法大多依据空间邻近域的非空间属性差异来计算偏离因子,忽略了邻近域内空间实体间距离的影响。本文首先讨论了空间邻近域内实体间距离对空间异常检测的影响,在此基础上,提出了一种顾及邻近域内实体间距离的空间异常度量方法——SOM法,并分析了它的复杂度。由于该方法是利用实体非空间属性的加权内插值与实测值的差值作为度量空间异常程度的参数,从而顾及了邻近域内所有实体相互间距离对非空间属性偏离的影响,并且克服了现有检测方法在不均匀分布空间实体集内寻找空间异常的缺陷。最后,通过一个实际算例验证了所提方法的可行性和正确性。
Spatial anomaly detection has become an important research content in spatial data mining and knowledge discovery. Spatial anomalies contain many unexpected knowledge. Most of the existing spatial anomaly detection methods calculate the deviation factor based on the non-spatial attribute differences of spatial neighboring regions, The influence of the distance between spatial entities in the neighboring domain. In this paper, we first discuss the effect of the distance between entities in the spatial neighborhood on the spatial anomaly detection. Based on this, we propose a method of spatial anomaly measurement that takes into account the distance between entities in a neighboring domain, the SOM method, and analyze its complexity. Because this method uses the difference between the weighted interpolation and the measured value of the non-spatial attribute of the entity as the parameter of measuring the degree of spatial anomaly, the influence of the distance between all the entities in the neighboring domain on the non-spatial attribute deviation is taken into account and the existing The detection method seeks for the defect of spatial anomaly in the heterogeneous spatial entity set. Finally, a practical example is used to verify the feasibility and correctness of the proposed method.