中文摘要：

针对犯罪空间聚集态研究算法中有关窗宽优化选择问题,将动态优化窗宽算法与DENCLUE（DENsity-based CLUstEring）算法结合,提出一种新的聚类算法。以入室盗窃案件为例,研究了该算法在犯罪热点探测方面的应用。结果显示引入优化窗宽算法后,可得到较为精确的聚类中心位置及概率密度变化趋势,并且当格网边长与邻近格网距离阈值及邻近点距离阈值之比为2：3：1时会得到较好的热点分析结果。通过分析犯罪热点分布图,可协助公安机关调整警力配置,加强案件高发区警力巡逻。

英文摘要：

The optimal bandwidth choose is a key parameter in the crime spatial aggregation states analysis method. The bandwidth is selected by coupling DENCLUE （DENsity-based CLUstEring） with the dynamic optimization method to cluster data for detecting high crime areas. Tests on a number of crime data sets show that the method gives precise locations of the cluster centers and density variations. When the ratio of the element side, the neighbouring grid distance threshold and the neighbouring point distance threshold is 2 ： 3 ： 1, the algorithm gives the best results. Analysis of the distribution of crime hot spots can assist police in allocating police resources to these high-crime areas.