犯罪案件时空分布模型研究

Research on Spatio-temporal Distribution Model of Criminal Cases

以社区犯罪案件为主要对象,研究马鞍山市犯罪案件的空间分布特征。结合数据挖掘技术、GIS技术与数学建模,根据犯罪心理学和刑侦学相关理论,建立OLR、GWR和GTWR 3种犯罪地理目标模型,对犯罪活动进行模型拟合。通过对比研究发现,GTWR模型考虑了案件的时间与空间特征,相较于最小二乘回归与地理加权回归在时空平稳性、均方根误差、平均绝对误差、平均绝对百分误差和模型拟合精度都得到了较好的效果。

In this paper,we mainly studied the community crime cases and the spatial distribution characteristics of crime cases in Maanshan City.Combining with the data mining technology,GIS technology and mathematical modeling,according to the theory of criminal psychology and criminal investigation,we established three kinds of criminal geographic target models of OLR,GWR and GTWR,and model fitted the criminal activities.The GTWR model introduced the spatio-temporal characteristics of the case and performed least squares regression.Comparing with least squares regression and geographically weighted regression,the spatio-temporal stability,root mean square error,mean absolute error,absolute percent error and model fitting accuracy of the GTWR model have achieved,good results.