Research On The Choice And Prediction Of Spatial Behavior Of Theft Criminals

Previous studies have shown that criminal travel behavior is affected by the individual characteristics of the criminal group,the spatial environment of the crime location,and different time periods.At present,there have been relatively abundant researches in the field of criminal travel research at home and abroad,but there are still some deficiencies in the description of the travel behavior of criminal groups.For example,for the calculation of spatial travel distance of criminal groups,the Manhattan distance is mainly used the approximation method of the criminal group,and the lack of in-depth exploration of the social characteristics of the criminal group,such as organization and group,in terms of the individual characteristics of the criminal group.Therefore,in view of the shortcomings in the calculation of the spatial travel distance of criminal groups and the analysis of influencing factors,based on the data of burglary cases in Beijing from 2015 to 2017,the travel distance of different criminal groups and its influencing factors were systematically studied.Using the Baidu map open platform to calculate and compare the European data,Manhattan distance,walking distance,cycling distance,driving distance and bus distance between the criminal group’s temporary residence and the crime scene,six travel distances,and combined with the criminal group The spatial element characteristics,time characteristics and individual characteristics of the criminal group of the temporary residence and the crime scene were analyzed,and the factors and variables affecting the spatial travel distance of the criminal group were analyzed.The incident was predicted.The analysis shows that the walking distance between the temporary residence of the criminal group and the crime site is the shortest,and the bus distance is the largest,but there is no obvious difference in the influence of spatial elements,time and individual characteristics of the criminal group,which proves that the space The distance measurement has no obvious effect on the travel behavior of the criminal group;the temporary residence of the criminal group and the distribution of spatial elements around the crime site have a significant impact on their spatial travel.The higher the road accessibility and the stronger the hot spot effect,the easier it is to attract criminal groups;in terms of time characteristics,the spatial travel distance of criminal groups is lower at the beginning of each year and the beginning of each month,and higher on rest days.,higher in the afternoon and evening within a day;in terms of the individual characteristics of the criminal group,there are obvious differences in the spatial travel distance of different criminal groups,among which,the criminal groups with characteristics such as male,middle-aged and group have a longer travel distance.By comparing the prediction models of various classification algorithms,the prediction effect of ridge regression can reach an accuracy higher than 80%,which is better than KNN,Naive Bayes,stochastic gradient descent classifier,support vector machine,multilayer perceptron,etc.algorithm.The work of this paper further enriches and supplements the empirical research in the field of crime geography in China,and has certain value for the practical work of public security in the fields of investigation,prevention and control.