基于随机森林和时空核密度方法的不同周期犯罪热点预测对比

doi:10.18306/dlkxjz.2018.06.003

地理科学进展 ›› 2018, Vol. 37 ›› Issue (6): 761-771.doi: 10.18306/dlkxjz.2018.06.003

基于随机森林和时空核密度方法的不同周期犯罪热点预测对比

柳林^1,^2,³(), 刘文娟¹, 廖薇薇¹, 余洪杰¹, 姜超¹, 林荣平¹, 纪佳楷¹, 张政¹

1. 中山大学地理科学与规划学院综合地理信息研究中心,广州 510275
2. 广州大学地理科学学院公共安全地理信息分析中心,广州 510006
3. 辛辛那提大学地理系,美国辛辛那提 OH45221-0131

收稿日期:2018-02-02 修回日期:2018-03-29 出版日期:2018-06-28 发布日期:2018-06-28
作者简介:
作者简介：柳林(1965-)，男，湖南湘潭人，博士，教授，主要研究方向为地理信息科学、犯罪时空分析与模拟等，E-mail: lin.liu@uc.edu。
基金资助:
国家自然科学重点基金项目(41531178)；广东省自然科学基金研究团队项目(2014A030312010)；国家自然科学基金项目(41171140)；广东省科技计划项目(2015A020217003)

Comparison of random forest algorithm and space-time kernel density mapping for crime hotspot prediction

Lin LIU^1,^2,³(), Wenjuan LIU¹, Weiwei LIAO¹, Hongjie YU¹, Chao JIANG¹, Rongping LIN¹, Jiakai JI¹, Zheng ZHANG¹

1. Center of Integrated Geographic Information Analysis, School of Geography and Planning, Sun Yat-Sen University, Guangzhou 510275, China
2. Center of Geographic Information Analysis for Public Security, School of Geographic Sciences, Guangzhou University, Guangzhou 510006, China
3. Department of Geography, University of Cincinnati, Cincinnati OH45221-0131, Ohio, USA

Received:2018-02-02 Revised:2018-03-29 Online:2018-06-28 Published:2018-06-28
Supported by:
Key Program of National Natural Science Foundation of China, No.41531178; Research Team Program of Natural Science Foundation of Guangdong Province, China, No.2014A030312010; National Natural Science Foundation of China, No.41171140; Science and Technology Program of Guangdong Province, China, No.2015A020217003

摘要/Abstract

摘要：

犯罪预测对于制定警务策略、实施犯罪防控具有重要意义。机器学习和核密度是2类主流犯罪热点预测方法,然而目前还鲜有研究对这2类方法在不同时间周期下的犯罪预测效果进行系统比较,本文试图对此进行补充。本文以2013-2016年5月的公共盗窃犯罪历史数据作为输入,分别对比了在接下来2周、1个月、2个月、3个月4个不同时间周期随机森林方法与基于时空邻近性的核密度方法的犯罪热点预测效果,结果发现：在各时间周期上,随机森林分类热点预测方法的面积和案件量命中率均比时空核密度方法准确性高;并且2种方法均能有效地识别犯罪热点中的高发区域,其中在较小范围较短时间内随机森林识别热点中的高发区效率更高,而在较大范围较长时间周期上时空核密度方法识别高发区更优。

关键词: 时空核密度, 随机森林算法, 犯罪热点预测, 犯罪高发区识别

Abstract:

Crime prediction is of great significance for the formulation of police tactics and the implementation of crime prevention and control in different time periods. Machine learning and density mapping are two common approaches for crime hotspot prediction. However, there exists few published work that systematically compares the predicted results of these two approaches. This study aimed to fill the gap. With crime patterns uncovered from 2013 to May 2016, we predicted hot-spot distribution of theft crimes in the period of first two weeks of June, July, and August in 2016 by random forest algorithm and traditional space-time kernel density method and compared the two sets of predictions. The research area was divided into grid cells of 50 m×50 m. Each cell was predicted as either hot-spot or non-hot-spot area in the next predicting period. Then we overlaid the forecast results and location of real cases to evaluate the accuracy of the two methods. The results show that both the hit rate of area and cases of the random forest classification hot-spot prediction method are higher than that of the space-time kernel density within different periods. Both methods can effectively identify high-crime areas of crime hot spots in prediction. In a relatively short period of time and small area, the random forest classification hotspot prediction method is more effective than the space-time kernel density method. However, in a relatively long term and large area, the space-time kernel density crime risk estimation method yields better result in identifying high crime areas.

Key words: space-time kernel density, random forest algorithm, crime hotspot prediction, high crime areas identification

柳林, 刘文娟, 廖薇薇, 余洪杰, 姜超, 林荣平, 纪佳楷, 张政. 基于随机森林和时空核密度方法的不同周期犯罪热点预测对比[J]. 地理科学进展, 2018, 37(6): 761-771.

Lin LIU, Wenjuan LIU, Weiwei LIAO, Hongjie YU, Chao JIANG, Rongping LIN, Jiakai JI, Zheng ZHANG. Comparison of random forest algorithm and space-time kernel density mapping for crime hotspot prediction[J]. PROGRESS IN GEOGRAPHY, 2018, 37(6): 761-771.

图/表 10

图1

图2

图3

图4

表1

表2

图5

图6

图7

图8

参考文献 33

[17]	Chainey S, Reid S, Stuart N.2002. When is a hotspot a hotspot? A procedure for creating statistically robust hotspot maps of crime[A]//Kidner D, Higgs G, White S. Innovations in GIS 9: Socio-economic applications of geographic information science. London, UK: Taylor and Francis: 21-36.
[18]	Chainey S, Tompson L, Uhlig S.2008. The utility of hotspot mapping for predicting spatial patterns of crime[J]. Security Journal, 21(1-2): 4-28. doi: 10.1057/palgrave.sj.8350066
[19]	Clarke R V G, Eck J E.2005. Crime analysis for problem solvers in 60 small steps[J]. Washington, DC: Center for Problem Oriented Policing.
[20]	Clarke R V G, Felson M.2004. Routine activity and rational choice[M]. New Brunswick, NJ: Transaction Publishers.
[21]	Cohen L E, Felson M.1979. Social change and crime rate trends: A routine activity approach[J]. American Sociological Review, 44(4): 588-608. doi: 10.2307/2094589
[22]	Genuer R, Poggi J -M, Tuleau-Malot C.2010. Variable selection using random forests[J]. Pattern Recognition Letters,31(14): 2225-2236. doi: 10.1016/j.patrec.2010.03.014
[23]	Gorr W L, Harries R.2003. Introduction to crime forecasting[J]. International Journal of Forecasting, 19(4): 551-555. doi: 10.1016/S0169-2070(03)00089-X
[24]	Groff E R, Vigne N G L.2002. Forecasting the future of predictive crime mapping[J]. Analysis for Crime Prevention,29-57.
[25]	Grubesic T H, Mack E A.2008. Spatial-temporal interaction of urban crime[J]. Journal of Quantitative Criminology, 24(3): 285-306. doi: 10.1007/s10940-008-9047-5
[26]	Hirschfield A, Bowers K.2001. Mapping and analyzing crime data: Lessons from research and practice[J]. Minerva Medica, 63(49): 2736-40. doi: 10.1111/1467-9787.00274
[1]	陈鹏, 疏学明, 袁宏永, 等. 2011. 时空犯罪热点预测模型研究[J]. 系统仿真学报, 23(9): 1782-1786.
	[Chen P, Shu X M, Yuan H Y, et al.2011. Research about spatial-temporal forecasting of crime hotspots[J]. Journal of System Simulation, 23(9): 1782-1786.]
[27]	Kandaswamy K K, Chou K C, Martinetz T, et al.2011. AFPPred: A random forest approach for predicting antifreeze proteins from sequence- derived properties[J]. Journal of Theoretical Biology, 270(1): 56-62. doi: 10.1016/j.jtbi.2010.10.037 pmid: 21056045
[28]	Kianmehr K, Alhajj R.2008. Effectiveness of support vector machine for crime hot-spots prediction[J]. Applied Artificial Intelligence, 22(5): 433-458. doi: 10.1080/08839510802028405
[2]	李卫红, 闻磊, 陈业滨. 2017. 改进的GA-BP神经网络模型在财产犯罪预测中的应用[J]. 武汉大学学报: 信息科学版, 42(8): 1110-1116. doi: 10.13203/j.whugis20160911
	[Li W H, Wen L, Chen Y B.2017. Property crime forecast based on improved GA-BP Neural Network Model[J]. Geomatics and Information Science of Wuhan University, 42(8): 1110-1116.] doi: 10.13203/j.whugis20160911
[29]	Ratcliffe J H.2004. Crime mapping and the training needs of law enforcement[J]. European Journal on Criminal Policy and Research, 10(1): 65-83.
[30]	Ratcliffe J H.2006. A temporal constraint theory to explain opportunity-based spatial offending patterns[J]. Journal of Research in Crime and Delinquency, 43(3): 261-291. doi: 10.1177/0022427806286566
[3]	梁晓军, 高展. 2001. “十五”期间犯罪预测与侦查工作机制创新[J]. 公安研究, (7): 30-32.
	[Liang X J, Gao Z.2001. "Shiwu" qijian fanzui yuce yu zhencha gongzuo jizhi chuangxin[J]. Policing Studies, (7): 30-32.]
[31]	Rodriguez Galiano V F, Ghimire B, Rogan J, et al.2012. An assessment of the effectiveness of a random forest classifier for land- cover classification[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 67: 93-104. doi: 10.1016/j.isprsjprs.2011.11.002
[32]	Rummens A, Hardyns W, Pauwels L.2017. The use of predictive analysis in spatiotemporal crime forecasting: Building and testing a model in an urban context[J]. Applied Geography, 86: 255-261. doi: 10.1016/j.apgeog.2017.06.011
[4]	刘大千, 修春亮. 2012. 国内外犯罪地理学研究进展评析[J]. 人文地理, 27(2): 38-44.
	[Liu D Q, Xiu C L.2012. Review of studies on criminal geography[J]. Human Geography, 27(2): 38-44.]
[5]	王发曾. 1992. 城市犯罪发展趋势预测[J]. 城市问题, (6): 15-18.
	[Wang F Z.1992. Chengshi fanzui fazhan qushi yuce[J]. Urban Problems, (6): 15-18.]
[6]	徐冲, 柳林, 周素红. 2016. 基于临近相似性考虑的犯罪热点密度图预测准确性比较: 以DP半岛街头抢劫犯罪为例[J]. 地理科学, 36(1): 55-62. doi: 10.1329/j.cnki.sgs.2016.01.007
	[Xu C, Liu L, Zhou S H.2016. The comparison of predictive accuracy of crime hotspot density maps with the consideration of the near similarity:A case study of robberies at DP Peninsula[J]. Scientia Geographica Sinica, 36(1): 55-62.] doi: 10.1329/j.cnki.sgs.2016.01.007
[7]	徐冲, 柳林, 周素红, 等. 2013. DP半岛街头抢劫犯罪案件热点时空模式[J]. 地理学报, 68(12): 1714-1723.
	[Xu C, Liu L, Zhou S H.2013. The spatio-temporal patterns of street robbery in DP Peninsula[J]. Acta Geographica Sinica, 68(12): 1714-1723.]
[8]	阎耀军, 张明. 2013. 犯罪预测时空定位信息管理系统的构建[J]. 中国人民公安大学学报: 社会科学版, 29(4): 73-80.
	[Yan Y J, Zhang M.2013. fanzui yuce shikong dingwei xinxi guanli xitong de goujian[J]. Journal of People's Public Security University of China: Social Sciences Edition, 29(4): 73-80.]
[9]	Bogomolov A, Lepri B, Staiano J, et al.2014. Once upon a crime: towards crime prediction from demographics and mobile data[C]//International conference on multimodal interaction. ACM, 427-434.
[10]	Bowers K J, Johnson S D.2005. Domestic burglary repeats and space-time clusters[J]. European Journal of Criminology, 2(1): 67-92. doi: 10.1177/1477370805048631
[11]	Bowers K J, Johnson S D, Pease K.2004. Prospective hot-spotting: The future of crime mapping[J]. British Journal of Criminology, 44(5): 641-658 doi: 10.1093/bjc/azh036
[12]	Brantingham P L, Brantingham P J.1999. A theoretical model of crime hot spot generation[J]. Studies on Crime & Crime Prevention, 8(1): 7-26.
[13]	Brantingham P L, Brantingham P J.2008. The rules of crime pattern theory[M]. Devon, UK: Willan Publishing.
[14]	Breiman L.2001. Random forests[J]. Machine Learning, 45(1): 5-32. doi: 10.1023/A:1010933404324
[15]	Caplan J M, Kennedy L W, Miller J.2011. Risk terrain modeling: Brokering criminological theory and GIS methods for crime forecasting[J]. Justice Quarterly, 28(2): 360-381. doi: 10.1080/07418825.2010.486037
[16]	Chainey S, Ratcliffe J.2013. GIS and Crime Mapping[M]. London, UK: John Wiley & Sons.
[33]	Weisburd D.2015. The law of crime concentration and the criminology of place[J]. Criminology, 53(2): 133-157. doi: 10.1111/1745-9125.12070

	热点网格数/个	总案件量/起	单个网格最大案件量	热点网格案件量中位数	热点网格案件量平均值
2016年6月份前2周	72	76	2	1	1.05
2016年6月份	140	164	6	1	1.17
2016年6、7月份	282	350	12	1	1.24
2016年6、7、8月份	415	554	20	1	1.33

	周期性	临近性
输入训练数据	2013年5月份案件量	2016年1月份案件核密度
	2014年5月份案件量	2016年2月份案件核密度
	2015年5月份案件量	2016年3月份案件核密度
	/	2016年4月份案件核密度
输入测试数据	2013年6月份案件量	2016年2月份案件核密度
	2014年6月份案件量	2016年3月份案件核密度
	2015年6月份案件量	2016年4月份案件核密度
	/	2016年5月份案件核密度
输入分类标签	2016年5月份热点/非热点	/
输出分类标签	/	2016年6月份热点/非热点

基于随机森林和时空核密度方法的不同周期犯罪热点预测对比

Comparison of random forest algorithm and space-time kernel density mapping for crime hotspot prediction

RichHTML

PDF (PC)

赞

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 33

相关文章 0

Metrics

本文评价

推荐阅读 0