社会安全事件空间分布研究

在讨论社会安全事件空间分布特点的基础上,简要介绍用于空间自相关性分析的主要算法。重点对基于密度聚类(DENCLUE,density-based clustering)的算法的基本原理、过程及利用积分均方误差MISE进行精度检验等方面进行阐述。将该算法用于社会安全事件空间分布研究领域中的犯罪热点区域探测。分析结果表明,与以往分析方法相比DENCLUE算法具有抗噪声能力强、能生成任意形状的聚类等特点,可有效地用于社会安全事件空间分布区域的研究,但也发现算法在格网边长的确定及窗宽长度选择方面的不足,有待进一步研究。

Research on Spatial Distribution of Social Security Incidents

On the base of discussing the spatial distribution characters of social security incidents,a brief introduction was firstly made on the different methods of spatial auto-correlation analysis.Then,the DENCLUE(density-based clustering) method was expounded including the principle,analysis process and accuracy test based on MISE.Finally,this method was applied to the hot-spots detection of crime in the field of the spatial distribution of social security incidents.The results show that DENCLUE method has strong anti-noise,and can generate any shape of the cluster,so it can be used to the research on spatial distribution of the social security incidents efficiently.At the same time,there are also obvious flaws in the aspects of the determination of grid side length and the choice of width and length of grids,which needs further study.