Analysis Of Spatial-temporal Characteristics And Causes Of Theft

As the application of GIS in crime analysis becomes more and more extensive,the focus of crime research gradually shifts to the deep-seated reasons behind crime and crime prediction,which brings great help to the early warning of crime risks.Therefore,this article analyzes the distribution characteristics and laws of hotspot crimes through the combination of association rules and spatiotemporal hotspot matrix,combined with environmental criminology theory and psychological theory,and analyzes the causes behind thefts.Forecast the area where the theft occurred.This article takes a certain area of a city as a research area,and attempts to analyze the spatial distribution of theft cases,the combination of spatiotemporal correlation and spatiotemporal hotspot matrix on the four time scales of year,quarter,month,week and two spatial scales such as street and community.The cause analysis of hot spots and the prediction of the occurrence community of extreme gradient decision tree method.The main research contents of this article are as follows:(1)Taking theft cases as the research object,the global and local spatial autocorrelation test,kernel density cluster analysis and standard deviation ellipse trend analysis of the theft cases in the study area from multiple time scales have proved that the study area The internal theft cases have obvious characteristics of time and space differentiation.First,in terms of time distribution,the year and quarter scales show obvious characteristics of "low first,then high,high and low staggered";on the month scale,there is a clear "high and low staggered" characteristic of low peak period and peak period;on the week scale,there is a characteristic of "highest first,then low" on the day,and the frequency of committing crimes on working days is higher than on weekends.Secondly,in terms of spatial distribution,the cases are distributed from north to south,mainly concentrated in the southeast,and the streets such as Hai * Street are the main gathering places.The hot spots of the case are stable on the year and quarter scales.On the weekday scale,the trend is shifting from northwest to southeast,with hotspot centers stably distributed in the central area of Hai * Street.(2)By adopting the combination of spatiotemporal correlation analysis and spatiotemporal hotspot matrix analysis,the spatiotemporal characteristics of theft cases and the mechanism of theft formation are summarized.First of all,the genetic algorithm based on simulated annealing is applied to the analysis of spatiotemporal correlation of theft,and at the same time,the theft analysis is related to the normal time and place of people's work and rest on the street scale,and the cross mutation method of the algorithm reduces the time complexity and improves theefficiency of rule extraction,and it is found that the hot spots in the research area are mostly distributed in the public area during working hours.Secondly,the age and gender of the cooperating case and the type of theft were analyzed on the four streets with a high frequency of incidents selected by the spatial and temporal association rules,and the distribution patterns of thefts in different regions were summarized,that is,the main locations of the crimes were concentrated.In universities,shopping and entertainment areas,subway stations and intersections,the perpetrators are mainly between the ages of 18 and 33 years old,mostly males.The frequency of common theft and theft of three-car cases with vehicles as the main target is relatively frequent high.Finally,analyze the distribution characteristics of hot spots combined with environmental criminology theory and psychological theory,and summarize the causes of the hot spots of theft.That is,the psychological influence factors of theft are mainly composed of financial needs,desire-driven,and self-esteem maintenance.It is composed of three special points,which are crowded and complicated,such as lack of monitoring and whether it is in a geographical location that is easy to escape.(3)Summarize the influencing factors of spatio-temporal features through the above causes of theft hotspots,and divide them into different data sets according to the characteristics of each factor,and use extreme gradient decision tree models to predict the occurrence of theft at different time scales at the community scale,and the results compared with the commonly used svm model,it is proved that the prediction accuracy and accuracy of the model is higher than that of the svm model.Through the relevant indicators,it is proved that the prediction effect of the D3 data set combined by the characteristics of the number of neighboring universities,companies,parking lots and the number of cases in the previous year on the scale of the year is good;In the above,the D4 data set combined with the characteristics of common public areas such as companies,hospitals,research institutes,universities,shopping malls,and parking lots is better.At the same time,it proves the reliability of the extreme gradient decision tree model used in this paper,and this method based on the analysis of theft space-time cause analysis and community occurrence prediction can play a great role in the future theft warning.