没有合适的资源?快使用搜索试试~ 我知道了~
资源推荐
资源详情
资源评论
Geographic Profiling: Nowhere To Hide
Summary
The study of serial crimes prediction has been a fruitful area of mathematical
research for decades. Here we attempt to analyze and model the serial crimes
prediction based on the time and locations of the past crime scenes.
Firstly, we divide the serial crimes prediction problem into two small prob-
lems: one is to develop different schemes to generate a geographical profile and
then develop a technique to combine the results; the other is to generate a use-
ful prediction of the location of the next crime. In addition, we study the Peter
Sutcliffe’s murder case to illustrate every step of our solution.
For the first problem, based on the least effort principle, we build the Cen-
ter of Minimum Distance Model and obtain the possible residence area of the
criminal. Then based on Bayes’ Theorem and Rayleigh distribution function,
we build the Bayesian Model and get another possible residence area. After
that, we design a technique of using a minimum circle to cover the two areas to
combine the two schemes.
For the second problem, using Rayleigh distribution function, we obtain a
preliminary probability distribution of the crime site based on the residence de-
termined in the first problem . Taking geographical character and the offender’s
geographical preference into account, we utilize cluster analysis to divide all
the crime sites into 4 zones. In these 4 zones, we construct 4 two-dimensional
normal distributions around the 4 circle centers with the standard deviations
being the radii of the circles. In view of the influence of the crime time, we add a
time factor to the preliminary distribution. As a result, the preliminary distribu-
tion is modulated by geographical and temporal factors, producing an ultimate
prediction, which is rather satisfactory after validation.
Finally, an executive summary is presented for law enforcement officers.
2010年美国国际大学生数学建模竞赛国际一等奖--张位春(07试点4),李臻(07试点2),邢照军(07国贸2)
Contents
Summary................................................................................... 1
1 Introduction ....................................................................... 3
1.1 Background ..................................................................... 3
1.2 Geographic Profiling.......................................................... 3
1.3 Objectives........................................................................ 3
2 Problem Analysis ............................................................... 4
2.1 Overall Strategy................................................................ 4
2.2 The Strategies for Problem I................................................. 4
2.3 The Strategy for Problem II ................................................. 4
3 Center of Minimum Distance Model.................................. 5
3.1 Notations ........................................................................ 5
3.2 Simplifying Assumptions.................................................... 6
3.3 Center of Minimum Distance Model ..................................... 6
3.4 Case Study....................................................................... 6
4 Bayesian Model.................................................................. 9
4.1 Notations and Definitions ................................................... 9
4.2 Simplifying Assumptions.................................................... 10
4.3 Bayesian Model ................................................................ 10
4.4 Case Study....................................................................... 12
4.5 The Discussion of α............................................................ 12
5 Combination of the Results of the Two Schemes................ 13
6 Prediction of the Next Crime.............................................. 14
6.1 Initial Prediction Method .................................................... 14
6.2 Developed Prediction Method ............................................. 15
7 Validation of Prediction...................................................... 20
8 Evaluating of Solutions ...................................................... 20
8.1 Strengths ......................................................................... 20
8.2 Weaknesses...................................................................... 21
8.3 Future Consideration ......................................................... 21
9 Executive Summary for the Chief of Police........................ 22
9.1 Preparatory Work.............................................................. 22
9.2 Overview of the Potential Issues........................................... 22
9.3 Suggestion....................................................................... 23
References ................................................................................. 24
2
Team #7120 page3 of 24
1 Introduction
1.1 Background
Serial crimes have a serious effect on people’s lives and continuously chal-
lenge people’s baseline of morality. Thus, how to arrest the criminals before he
or she commits another crime becomes an important issue. A number of sophis-
ticated techniques have been developed to generate a useful prediction for law
enforcement officers based on the time and locations of the past crime scenes,
such as the “center of a mass” method.
It is well-known that the focus of any police investigation is the crime scene
and its evidentiary contents. What is often overlooked, however, is a geographic
perspective on the actions preceding the offense: the spatial behavior that leads
to the crime scene. For any violent crime to occur there must have been an in-
tersection in both time and place between the victim and offender[1]. Thus, D.
Kim Rossmo advanced a new criminal investigative methodology called “Geo-
graphic Profiling” in 2000[2].
1.2 Geographic Profiling
Geographic profiling is a criminal investigative methodology that analyzes
the locations of a connected series of crimes to determine the most probable area
of offender residence. Typically used in cases of serial murder or rape (but also
arson, bombing, robbery, and other crimes), the technique helps police detec-
tives prioritize information in large-scale major crime investigations that often
involve hundreds or thousands of suspects and tips [3].
1.3 Objectives
Our goal is to develop at least two different schemes to generate a geographi-
cal profile. After that, we are asked to develop a technique to combine the results
of the different schemes and generate a useful prediction. Our prediction should
provide some kind of estimate or guidance about possible locations of the next
crime based on the time and locations of the past crime scenes. What’s more, we
should show how reliable the estimate will be in a given situation according to
our model.
3
Team #7120 page4 of 24
2 Problem Analysis
2.1 Overall Strategy
For simplicity, we divide the problem into two small problems
• Problem I: To develop at least two different schemes to generate a geo-
graphical profile and then develop a technique to combine the results of
the different schemes.
• Problem II: To generate a useful prediction based on the geographical pro-
file.
Having ensured this, our strategy becomes
• Designing at least two different schemes to search out the residence of the
criminal firstly based on the time and locations of the past crime scenes
and designing a technique to combine the results of the different schemes.
• Utilizing the residence to predict the location of the next crime.
2.2 The Strategies for Problem I
After literature study, there are two categories of strategies to solve the Prob-
lem I: one is spatial distribution strategies and the other is probability distance
strategies[4].
Spatial distribution strategies include a number of different procedures, all of
which predict the home location of a serial offender by calculating a central point
from a distribution of crime site locations. Some common spatial distribution
strategies include the center of the circle, centroid, median, geometric mean,
harmonic mean, and center of minimum distance.
Probability distance strategies begin with the assumption that an offender’s
crime site locations define their activity space, and that this area contains the
offenders residence. Probability distance strategies differ from one another in
terms of the shape of the mathematical function applied around each crime site
and the assumptions regarding the relationship between where offenders reside
and where they commit their offences. Common probability distance functions
include the negative exponential, normal, lognormal, linear and truncated neg-
ative exponential.
The results of the schemes should be combined by certain technique to pro-
duce some useful geographical profile.
2.3 The Strategy for Problem II
It is well known that the next crime is committed at a place where the proba-
bility of crime is the highest. So, for prediction, we try to obtain a planar proba-
bility distribution of being the next crime site. To accomplish this, several factors
4
Team #7120 page5 of 24
should be taken into consideration, including geographical character, the crimi-
nal’s mentality, the time of previous crimes, and so forth.
3 Center of Minimum Distance Model
3.1 Notations
Tab 1: Notations
Name Description
s the total number of crimes
x
i
the abscissa of the i
th
crime site(i = 1, 2,· · · ,s)
y
i
the ordinate of the i
th
crime site(i = 1, 2,· · · ,s)
z the location of the criminal’s residence
m the abscissa of the criminal’s residence
n the ordinate of the criminal’s residence
D
i
(z) the distance between the criminal’s residence and the i
th
crime
site(i = 1, 2,· · · ,s)
In order to illustrate the variables more clearly, we draw the following sketch.
Fig 1: The sketch of the crime site(s = 6)
5
剩余23页未读,继续阅读
资源评论
喜多
- 粉丝: 8
- 资源: 30
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功