基于数据挖掘的社区犯罪率分析与预测研究

发布时间：2018-06-23 08:46

本文选题：数据挖掘 + 社区犯罪率　；参考：《北京交通大学》2017年硕士论文

【摘要】：社区是人们生活的基本单元,与每个人的生活息息相关,在犯罪手段和方式不断翻新的今天,怎样对社区犯罪进行预警分析具有重大的意义。目前公安系统及相关行业都积累了海量数据,这些都为数据挖掘创造了前提条件。本文利用数据挖掘的方法,以社区数据及相关犯罪数据为研究对象,针对社区犯罪率分析及预测这一目的,建立了社区犯罪率分析及预测体系,揭示出社区犯罪率与其影响因素之间的内在规律性,得出影响社区犯罪率的关键因素,并对社区潜在犯罪率进行预测,为警方合理部署有限资源与重点排查提供科学依据。本文的主要研究内容如下:1、建立基于数据挖掘的社区犯罪率分析与预测体系。选用k-means聚类的方法将社区分成了犯罪率高、中、低三类,横向对比具有不同犯罪率的社区之间的属性差异,并通过岭回归与lasso回归得出了影响社区犯罪率的关键因素,最后使用支持向量机对社区潜在犯罪率进行预测,为合理评估社区的安全性提供技术及理论支持。2、多元社区数据与犯罪数据的处理。针对社区数据与相关犯罪数据具有数量较大且多元化的特点,采用了拉格朗日插值法、z-score数据标准化方法以及主成分分析法(PCA)对社区及犯罪数据进行了预处理,为挖掘工作的有序进行创造了前提条件。3、社区犯罪率分析与预测体系的应用。为了验证本文提出的社区犯罪率分析与预测体系的可行性,结合一组真实的社区数据与相关的犯罪数据实现犯罪率分析与预测体系的实例分析,取得了较好的分类及预测效果,筛选出了影响社区谋杀案件犯罪率的权重最大的几个因素,并进行社区潜在犯罪率的预测,因此可以使用本文提出的评价体系来评估社区的安全性指数。4、根据社区犯罪数量与社区人均收入、警察数量的数据图,建立了三者之间的关系式,并通过实验拟合得到均衡解,从而量化社区警力资源分布,为犯罪预防工作提供决策支持。
[Abstract]:Community is the basic unit of people's life, which is closely related to everyone's life. At present, the public security system and related industries have accumulated massive data, which have created a prerequisite for data mining. Based on the method of data mining and taking community data and related crime data as the research object, this paper establishes a community crime rate analysis and prediction system for the purpose of community crime rate analysis and prediction. This paper reveals the inherent regularity between the crime rate in the community and its influencing factors, obtains the key factors affecting the crime rate in the community, and forecasts the potential crime rate in the community, which provides a scientific basis for the police to reasonably deploy the limited resources and focus on the investigation. The main contents of this paper are as follows: 1. A community crime rate analysis and prediction system based on data mining is established. The k-means clustering method is used to divide the community into three groups: high, middle and low crime rates. The attribute differences among communities with different crime rates are compared horizontally, and the key factors influencing the crime rate in communities are obtained by ridge regression and lasso regression. Finally, support vector machine is used to predict the potential crime rate in the community, which provides technical and theoretical support for the reasonable evaluation of community security. In view of the large quantity and diversity of community data and related crime data, the Lagrange interpolation method and principal component analysis (PCA) are used to preprocess community and crime data. It creates the precondition. 3. The application of community crime rate analysis and prediction system. In order to verify the feasibility of the community crime rate analysis and prediction system proposed in this paper, combining a group of real community data and related crime data to realize the crime rate analysis and prediction system, a case study has been carried out, and good classification and prediction results have been obtained. Selected the most important factors that influence the crime rate of murder cases in the community, and predicted the potential crime rate in the community. Therefore, the evaluation system proposed in this paper can be used to evaluate the community security index .4.According to the data graph of community crime and community per capita income, police number, the relationship between them is established, and the equilibrium solution is obtained by experiment fitting. So as to quantify the distribution of community police resources and provide decision support for crime prevention.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;D917

【参考文献】