犯罪共生空间的类型识别及其特征分析
作者简介:柳林(1965-),男,湖南湘潭人,博士,教授,博导,主要从事人文地理信息科学、犯罪时空分析与模拟研究。E-mail: lin.liu@uc.edu
收稿日期: 2018-01-08
要求修回日期: 2018-05-25
网络出版日期: 2018-08-20
基金资助
国家重点研发计划项目(2018YFB0505500,2018YFB0505503)、国家自然科学基金重点项目(41531178)、广州市科学研究 计划重点项目(201804020016)、广东省自然科学基金研究团队项目(2014A030312010)资助
Detecting and Characterizing Symbiotic Clusters of Crime
Received date: 2018-01-08
Request revised date: 2018-05-25
Online published: 2018-08-20
Supported by
National Key R&D Program of China (2018YFB0505500, 2018YFB0505503), Key Program of National Natural Science Foundation of China (41531178), Key Project of Science and Technology Program of Guangzhou City, China (201804020016), Research Team Program of Natural Science Foundation of Guangdong Province, China (2014A030312010)
Copyright
以ZG市公共空间盗窃、入室盗窃、寻衅滋事、接触诈骗、抢劫抢夺犯罪为研究对象,采用K均值聚类法识别不同类型的犯罪共生空间;并通过决策树模型分析了不同犯罪共生空间的特征。结果表明,ZG市犯罪共生空间可划分为4种类型:① 无犯罪类型共生区;② 公共空间盗窃和接触诈骗犯罪共生区;③ 所有类型犯罪共生区;④ 入室盗窃、寻衅滋事和抢劫抢夺犯罪共生区。城市中各异的社会环境和建成环境产生了不同类型的犯罪机会,而且各类社会环境和建成环境之间存在着条件交互性作用。研究结果为制定犯罪的联合防控策略和实现有限警力的合理布控并且提高执法效率提供了理论基础。
柳林 , 杜方叶 , 宋广文 , 龙冬平 , 姜超 , 肖露子 . 犯罪共生空间的类型识别及其特征分析[J]. 地理科学, 2018 , 38(8) : 1199 -1209 . DOI: 10.13249/j.cnki.sgs.2018.08.001
Research on the spatial distribution of crime most often shows that the spatial distribution of crime is not homogeneous. Moreover, a subset of this literature shows that spatial distributions of different crime types show similarities across urban space. This raises the possibility of symbiotic relationships between different types of crime in urban space. Previous studies focus on spatial concentrations, spatial patterns and hotspot distributions of crime. They ignore associations between different types of crime in space. This paper arms to fill the gap by examining the association between different types of crime, detecting symbiotic clusters of crime, and characterizing these clusters. So theft, burglary, affray, fraud and robbery are extracted from the call for services data of ZG city. Application of the K-means clustering algorithm on these data detects symbiotic clusters of various types of crime. Points of interest from commercial navigation data sets and the sixth census data are used to characterize the socio-economic environments of the symbiotic clusters, with the assistance of the decision tree algorithm of Weka. The results show that ZG city can be divided into 4 symbiotic clusters of crime: 1) low incidence of all crime; 2) high incidence of theft, fraud and low incidence of burglary, affray, and robbery; 3) high incidence of all crime; 4) high incidence of burglary, affray, robbery and low incidence of theft, and fraud. Cluster 1 is characterized by high bus station density, low proportion of floating population, high proportion of elder and low catering facilities density. The social and physical environment of cluster 1 generate only a few convergences of motivated offenders, suitable targets, and incapable guardians. As such, all crime rates becomes low due to the lack of crime opportunities. Cluster 2 is characterized by high bus station density, high proportion of floating population, high proportion of young and low proportion of rental housing. The social and physical environment of cluster 2 host large volume of convergences of motivated offenders, suitable targets of and incapable guardians, generating extensive opportunities for larceny and fraud. Cluster 3 is characterized by high bus station density, high proportion of floating population, low proportion of young, high proportion of rental housing and high catering facilities. Ordinary crime emerges when a likely offender converges with a suitable crime target in the absence of a capable guardian against crime. The social and physical environment of cluster 3 can make such convergences much more likely for all types of crime. Cluster 4 is characterized by low bus station density, high catering facilities density, high supermarkets density, low proportion of young and high residential quarter density. The social and physical environment of cluster 3 bring down the number of potential offenders and targets, and subsequently the opportunities for burglary, affray and robbery. The potential value of the results is to provide useful guidance for the joint prevention of different type of crime.
Table 1 The percent of police district of top 50%, 80% of crime(N=176) and Lorentz coefficient表1 2014年ZG市占各类型犯罪警情50%、80%的派出所比例(N=176)及洛伦兹系数 |
犯罪类型 | 50%警情 | 80%警情 | 洛伦兹系数 |
---|---|---|---|
公共空间盗窃 | 22.7% | 50.6% | 0.949 |
入室盗窃 | 19.3% | 47.7% | 0.933 |
寻衅滋事 | 20.5% | 50.6% | 0.931 |
接触诈骗 | 22.7% | 51.7% | 0.953 |
抢劫抢夺 | 10.2% | 31.8% | 0.880 |
Fig.1 Spatial distribution of all types of crime in ZG City图1 2014年ZG市各类型犯罪空间分布格局 |
Table 2 Outcomes of hierarchical clustering and symbiotic clusters of crime表2 聚类分析结果及各类犯罪共生空间划分 |
类别 | 警情类型 | 极小值 | 极大值 | 均值 | 标准差 | 排名* |
---|---|---|---|---|---|---|
第一类(I) (N=38) | a公共空间盗窃 | 59 | 176 | 148.97 | 24.39 | 149 |
b入室盗窃 | 91 | 176 | 148.32 | 21.06 | 148 | |
c寻衅滋事 | 87 | 176 | 147.05 | 22.68 | 147 | |
d接触诈骗 | 77 | 176 | 147.31 | 23.69 | 147 | |
e抢劫抢夺 | 74 | 176 | 145.76 | 27.18 | 146 | |
第二类(II)(N=37) | a公共空间盗窃 | 5 | 126 | 66.57 | 30.93 | 67 |
b入室盗窃 | 77 | 166 | 122.84 | 25.45 | 123 | |
c寻衅滋事 | 28 | 169 | 107.35 | 42.78 | 107 | |
d接触诈骗 | 3 | 138 | 72.51 | 38.33 | 73 | |
e抢劫抢夺 | 56 | 168 | 114.03 | 29.52 | 114 | |
第三类(III)(N=49) | a公共空间盗窃 | 1 | 89 | 35.14 | 23.80 | 35 |
b入室盗窃 | 1 | 135 | 38.49 | 33.91 | 38 | |
c寻衅滋事 | 1 | 99 | 29.53 | 21.97 | 30 | |
d接触诈骗 | 1 | 94 | 36.41 | 24.18 | 36 | |
e抢劫抢夺 | 1 | 117 | 34.71 | 26.76 | 35 | |
第四类(IV)(N=52) | a公共空间盗窃 | 19 | 162 | 110.19 | 31.29 | 110 |
b入室盗窃 | 17 | 122 | 67.48 | 25.92 | 67 | |
c寻衅滋事 | 31 | 130 | 87.86 | 25.18 | 80 | |
d接触诈骗 | 25 | 164 | 105.98 | 35.09 | 106 | |
e抢劫抢夺 | 15 | 156 | 79.17 | 36.04 | 79 |
注:*排名越靠前代表犯罪率越高,排名越靠后代表犯罪率越低。即排名为1的派出所犯罪率最高,排名为176的派出所犯罪率最低。 |
Table 3 The results of PAI Index表3 PAI指数计算结果 |
类别 | 公共间 盗窃 | 入室 盗窃 | 寻衅 滋事 | 接触 诈骗 | 抢劫 抢夺 |
---|---|---|---|---|---|
第一类(I) | 0.243 | 0.205 | 0.248 | 0.292 | 0.116 |
第二类(II) | 4.417 | 2.236 | 2.782 | 4.438 | 1.364 |
第三类(III) | 2.082 | 2.216 | 2.262 | 1.914 | 2.875 |
第四类(IV) | 0.533 | 0.778 | 0.647 | 0.580 | 0.600 |
Fig.2 Spatial distribution of symbiotic space图2 犯罪共生空间分布 |
Fig.3 Abstraction of part of decision trees图3 部分决策树示意图 |
The authors have declared that no competing interests exist.
[1] |
[
|
[2] |
|
[3] |
|
[4] |
|
[5] |
|
[6] |
|
[7] |
|
[8] |
|
[9] |
|
[10] |
|
[11] |
|
[12] |
|
[13] |
|
[14] |
|
[15] |
[
|
[16] |
[
|
[17] |
|
[18] |
[
|
[19] |
[
|
[20] |
|
[21] |
|
[22] |
|
[23] |
|
[24] |
|
[25] |
[
|
[26] |
|
[27] |
[
|
[28] |
[
|
[29] |
[
|
[30] |
[
|
[31] |
|
[32] |
|
[33] |
|
[34] |
|
[35] |
|
[36] |
|
[37] |
|
[38] |
Sampson and
|
[39] |
|
[40] |
|
[41] |
|
[42] |
|
[43] |
|
[44] |
|
/
〈 |
|
〉 |