摘要
本文针对传统犯罪预测模型中数据处理方法不适用于犯罪近重复性分析,以及犯罪数据高度不平衡特性导致的犯罪预测结果分散的问题,提出了一种基于Mnd-Knox算法以及时空交互网格结构改进的犯罪预测模型.该模型框架由确定时空阈值、建立时空网格结构、扩充数据集、数据挖掘4部分组成.首先采用Mnd-Knox算法确定犯罪近重复性显著的时空域值;其次采用网格化地理信息管理方法建立网格结构,并确定各因子间影响权重;然后在基础数据集上融合附加地理环境特征;最后采用深度神经网络算法进行数据挖掘.针对2016年芝加哥地区4类频发型犯罪数据进行实验.结果表明,与传统犯罪预测模型相比,本文所提出的模型构建方法有更好的预测效果,模型平均绝对误差值降低了88.56%.
Prediction model based on traditional crime near-repeat data processing method is not applicable to crime analysis,and crime data highly unbalanced characteristics crime prediction results dispersion problem,put forward a kind of based on Mnd-Knox algorithm as well as the interaction of time and space grid structure improvement of crime prediction model.The framework of the model consists of four parts:determining the spatial and temporal threshold,establishing the spatio-temporal grid structure,expanding the data set,and data mining.Firstly,the Mnd-Knox algorithm is used to determine the spatio-temporal domain values with significant near-repeat of crimes.Secondly,the grid structure was established by the grid geographic information management method,and the influence weights among the factors were determined.Secondly,the additional geographical environment features are fused on the basis of the general data set.Finally,deep neural networks algorithm is used for data mining.An experiment was conducted on the data of four types of frequent crime records in Chicago,USA in 2016.The results show that compared with the traditional crime prediction model,the proposed model construction method has better prediction effect,and the average absolute error of the model is reduced by 88.56%.
作者
魏东
张天祎
WEI Dong;ZHANG Tian-yi(School of Electrical and Information Engineering,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;Beijing Key Laboratory of Intelligent Processing for Building Big Data,Beijing Municipal Science and Technology Commission,Beijing 100044,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2022年第11期2456-2464,共9页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(61871020)资助
北京市属高校高水平创新团队建设计划项目(IDHT20190506)资助
北京市教委科技计划重点项目(KZ201810016019)资助.
关键词
犯罪预测
特征优化
时空模型
犯罪自相关性
crime prediction
feature optimization
spatio-temporal model
crime autocorrelation
