SlideShare a Scribd company logo
1 of 2
Download to read offline
Reallocating Patrol Zones Based on Previous Crime Pattern
Kyung Kim, Industrial and Systems Engineering
Overview of Proposed Work
Campus safety is one aspect that should not be overlooked when students decide which college to attend.
For that reason, the Georgia Tech Police Department (GTPD) invested in new technologies and selected
brave officers to ensure safety on campus. A decrease in annual crime rate at Georgia Tech for past few
years shows that they have been producing fruitful results. However, our team wanted our campus to be
even more harmless. With this goal in mind, in Fall 2014, we asked the GTPD what they needed help with.
In our initial meeting with them, we summarized main topics we wanted to investigate on. These included
resource allocation, crime clustering, and crime prediction to name a few.
After a semester of work on those topics, we came up with some interesting results. We located clustered
crime locations in Georgia Tech and Atlanta with help of the k-means clustering algorithm, predicted the
number of crimes in the future using time series analysis, and found out that some crimes led to another crimes
(e.g. stolen cars were used to commit robbery). Also, we found inconsistency of number of crimes in each
zone to be the consequence of having inefficient patrol routes. However, we could fix this by incorporating
results we mentioned above and suggesting a new set of patrol zones.
Currently, Georgia Tech is divided into 4 zones. Among these zones, zone 2 has noticeably more crimes
than any other zones. This is a natural phenomenon, as zone 2 covers more scandalous regions than others do.
However, if we strategically define newly partitioned zones, then we will be able to suggest more reasonable
patrol routes. We can do this by integrating our prior findings. Our data on clustering could assist us to
rationally divide up the zones, and our forecasted crime locations found from time series analysis will let
us check adequacy of our new zones. Therefore, for PURA and for our safety on campus, we would like to
propose a new way to allocate GTPD patrol zones.
Objective and Goals for the Semester
Our primary goal is to make the Georgia Tech campus safer, and we can do it by increasing the police
patrol efficiency by redesigning patrol zones such that the crime occurrences in each zone become approx-
imately uniform. That way, the police officers have higher chance of encountering criminals and also can
easily go cover for colleagues. We hope that, at the end of this semester, our results not only help the police
department, but also can invoke interest from other students regarding campus safety so that more students
will be encouraged to participate in research with the GTPD. We will produce a satisfying result that will
make students realize what we can contribute to our campus safety. To meet our goals for this semester, we
plan to complete the tasks listed below.
1. Data mining
(a) Clean up data
(b) Select significant variables
2. Crime pattern analysis
(a) Find noticeable crime patterns
(b) Look for relationships among crimes
3. Time series analysis
(a) Predict future crimes
i. Location
ii. Time
iii. Type of crime
4. Reallocate patrol zones
(a) Make and solve optimization models
(b) Run simulation
(c) Suggest new patrol routes based on new
patrol zones
1
Methods and Techniques to be Used
Our main methodology for this task will include clustering algorithms, time series analysis, and big
data analysis. Some specific techniques we will apply are k-means clustering algorithms and Holt-Winters
smoothing method. For our research, we will use software such as Microsoft Excel, Minitab, R-studio, Arena,
Google Earth, and CrimeStat. We will use Microsoft Excel to manipulate data; Minitab and R-studio to
run statistical tests; Arena to run simulation; and Google Earth and CrimeStat to run algorithms and make
visual representations. Any coding will be done with Python, and it will be our main method of solving
optimization models. Furthermore, we may use PredPol, a crime predicting software, if we were given a
permission.
We have four years of crime data from 2011 to 2014, with more than 200 variables. We will first clean
up the data and remove statistically insignificant variables. Then, with the remaining data, we will apply
various techniques to see what we can learn from them. As our research will be directed to observe crime
patterns, areas of concentrated crimes (hot spots), and relationships among crimes, we will most likely apply
algorithms to check for clustering, perform time series methods to predict crimes for 2015 and 2016, and run
simulations on our model to verify appropriateness on our newly partitioned zones.
Many of our optimization model will be based on mixed integer programming. As we deal with future
events, we will apply robust optimization to deal with uncertainties. Our goal will be to minimize constraints
from unknown possibilities and to deal with worst case to base case scenarios.
When we run simulations, we will use discrete-event simulations and check for inter-arrival time, service
time, and response time because this provides a way of solving complex situations, such as non-constant
arriving rates [1]. We will run simulation several times, and after each run, we will make a few changes in
the model and see how much improvement we can make from that. Since we will run simulation numerous
times, our task will be to come up with an elegant algorithm that will closely mimic real life situations so
that we could solve the problem in minimum number of iterations.
Research and Background Experience
Our research team has been working with Professor David Goldberg on various areas of studies before
we settled on to this project. Our group has many strengths, but our greatest strength is that we are
specialized. For example, we have a member who is an expert at Microsoft Excel, another member who is
good at coding, and another who has a great knowledge of Statistics. I take part on working with theoretical
computer science and mathematical aspects of research. Furthermore, we have diverse working experience,
as we have worked for an airline industry, supply chain companies, and a technology company. Overall, we
are all motivated to work hard and to produce a rewarding result.
As mentioned in the overview, our research team has finished all the preliminary work [2], presented our
result to GTPD, and are now prepared to begin a new project that police officers consider to be valuable
for them.
References
[1] Zhang, Y., Brown, D.: Simulation Optimization of Police Patrol District Design Using an Adjusted
Simulated Annealing Approach, Department of Systems and Information Engineering
[2] Kim, K. Wang, D. Wang, Y. Bao, B. Park, S. Lee, H.: Georgia Tech and Atlanta Crime Analysis. (2015).
Link: https:/drive.google.com/file/d/0B9WaQKspypsVQWJKX1Nub29sT3c/view?usp=sharing
2

More Related Content

What's hot

AM 207_Poster Final
AM 207_Poster FinalAM 207_Poster Final
AM 207_Poster FinalYaxiong Cai
 
Master Thesis Presentation
Master Thesis PresentationMaster Thesis Presentation
Master Thesis PresentationEhab Qadah
 
Crime Pattern Detection using K-Means Clustering
Crime Pattern Detection using K-Means ClusteringCrime Pattern Detection using K-Means Clustering
Crime Pattern Detection using K-Means ClusteringReuben George
 
Performance evaluation of hypercube interconnectionm
Performance evaluation of hypercube interconnectionmPerformance evaluation of hypercube interconnectionm
Performance evaluation of hypercube interconnectionmAnjali Agrawal
 
Crime analysis mapping, intrusion detection using data mining
Crime analysis mapping, intrusion detection using data miningCrime analysis mapping, intrusion detection using data mining
Crime analysis mapping, intrusion detection using data miningVenkat Projects
 
Presentation iswc
Presentation iswcPresentation iswc
Presentation iswcSydGillani
 
Building and Measuring Privacy-Preserving Mobility Analytics
Building and Measuring Privacy-Preserving Mobility AnalyticsBuilding and Measuring Privacy-Preserving Mobility Analytics
Building and Measuring Privacy-Preserving Mobility AnalyticsEmiliano De Cristofaro
 

What's hot (8)

AM 207_Poster Final
AM 207_Poster FinalAM 207_Poster Final
AM 207_Poster Final
 
Master Thesis Presentation
Master Thesis PresentationMaster Thesis Presentation
Master Thesis Presentation
 
Crime Pattern Detection using K-Means Clustering
Crime Pattern Detection using K-Means ClusteringCrime Pattern Detection using K-Means Clustering
Crime Pattern Detection using K-Means Clustering
 
Predictive Policing
Predictive PolicingPredictive Policing
Predictive Policing
 
Performance evaluation of hypercube interconnectionm
Performance evaluation of hypercube interconnectionmPerformance evaluation of hypercube interconnectionm
Performance evaluation of hypercube interconnectionm
 
Crime analysis mapping, intrusion detection using data mining
Crime analysis mapping, intrusion detection using data miningCrime analysis mapping, intrusion detection using data mining
Crime analysis mapping, intrusion detection using data mining
 
Presentation iswc
Presentation iswcPresentation iswc
Presentation iswc
 
Building and Measuring Privacy-Preserving Mobility Analytics
Building and Measuring Privacy-Preserving Mobility AnalyticsBuilding and Measuring Privacy-Preserving Mobility Analytics
Building and Measuring Privacy-Preserving Mobility Analytics
 

Similar to Kyung Kim

Predictive Modeling for Topographical Analysis of Crime Rate
Predictive Modeling for Topographical Analysis of Crime RatePredictive Modeling for Topographical Analysis of Crime Rate
Predictive Modeling for Topographical Analysis of Crime RateIRJET Journal
 
Crime Data Analysis, Visualization and Prediction using Data Mining
Crime Data Analysis, Visualization and Prediction using Data MiningCrime Data Analysis, Visualization and Prediction using Data Mining
Crime Data Analysis, Visualization and Prediction using Data MiningAnavadya Shibu
 
CRIME ANALYSIS AND PREDICTION USING MACHINE LEARNING
CRIME ANALYSIS AND PREDICTION USING MACHINE LEARNINGCRIME ANALYSIS AND PREDICTION USING MACHINE LEARNING
CRIME ANALYSIS AND PREDICTION USING MACHINE LEARNINGIRJET Journal
 
Crime Dataset Analysis for City of Chicago
Crime Dataset Analysis for City of ChicagoCrime Dataset Analysis for City of Chicago
Crime Dataset Analysis for City of ChicagoStuti Deshpande
 
Predictive Modeling for Topographical Analysis of Crime Rate
Predictive Modeling for Topographical Analysis of Crime RatePredictive Modeling for Topographical Analysis of Crime Rate
Predictive Modeling for Topographical Analysis of Crime RateIRJET Journal
 
A Survey on Data Mining Techniques for Crime Hotspots Prediction
A Survey on Data Mining Techniques for Crime Hotspots PredictionA Survey on Data Mining Techniques for Crime Hotspots Prediction
A Survey on Data Mining Techniques for Crime Hotspots PredictionIJSRD
 
Crime Prediction and Analysis
Crime Prediction and AnalysisCrime Prediction and Analysis
Crime Prediction and AnalysisIRJET Journal
 
A predictive model for mapping crime using big data analytics
A predictive model for mapping crime using big data analyticsA predictive model for mapping crime using big data analytics
A predictive model for mapping crime using big data analyticseSAT Journals
 
IRJET - Crime Analysis and Prediction - by using DBSCAN Algorithm
IRJET -  	  Crime Analysis and Prediction - by using DBSCAN AlgorithmIRJET -  	  Crime Analysis and Prediction - by using DBSCAN Algorithm
IRJET - Crime Analysis and Prediction - by using DBSCAN AlgorithmIRJET Journal
 
Survey on Crime Interpretation and Forecasting Using Machine Learning
Survey on Crime Interpretation and Forecasting Using Machine LearningSurvey on Crime Interpretation and Forecasting Using Machine Learning
Survey on Crime Interpretation and Forecasting Using Machine LearningIRJET Journal
 
Life and science journal.pdf
Life and science journal.pdfLife and science journal.pdf
Life and science journal.pdfSarita30844
 
Proposed Effective Solution for Cybercrime Investigation in Myanmar
Proposed Effective Solution for Cybercrime Investigation in MyanmarProposed Effective Solution for Cybercrime Investigation in Myanmar
Proposed Effective Solution for Cybercrime Investigation in Myanmartheijes
 
Software Defect Trend Forecasting In Open Source Projects using A Univariate ...
Software Defect Trend Forecasting In Open Source Projects using A Univariate ...Software Defect Trend Forecasting In Open Source Projects using A Univariate ...
Software Defect Trend Forecasting In Open Source Projects using A Univariate ...CSCJournals
 
Knowledge and Data Engineering IEEE 2015 Projects
Knowledge and Data Engineering IEEE 2015 ProjectsKnowledge and Data Engineering IEEE 2015 Projects
Knowledge and Data Engineering IEEE 2015 ProjectsVijay Karan
 
IRJET- Cyber Crime Attack Prediction
IRJET- Cyber Crime Attack PredictionIRJET- Cyber Crime Attack Prediction
IRJET- Cyber Crime Attack PredictionIRJET Journal
 
Propose Data Mining AR-GA Model to Advance Crime analysis
Propose Data Mining AR-GA Model to Advance Crime analysisPropose Data Mining AR-GA Model to Advance Crime analysis
Propose Data Mining AR-GA Model to Advance Crime analysisIOSR Journals
 
CRIME EXPLORATION AND FORECAST
CRIME EXPLORATION AND FORECASTCRIME EXPLORATION AND FORECAST
CRIME EXPLORATION AND FORECASTIRJET Journal
 
Analysis of Crime Big Data using MapReduce
Analysis of Crime Big Data using MapReduceAnalysis of Crime Big Data using MapReduce
Analysis of Crime Big Data using MapReduceKaushik Rajan
 

Similar to Kyung Kim (20)

Predictive Modeling for Topographical Analysis of Crime Rate
Predictive Modeling for Topographical Analysis of Crime RatePredictive Modeling for Topographical Analysis of Crime Rate
Predictive Modeling for Topographical Analysis of Crime Rate
 
Crime Data Analysis, Visualization and Prediction using Data Mining
Crime Data Analysis, Visualization and Prediction using Data MiningCrime Data Analysis, Visualization and Prediction using Data Mining
Crime Data Analysis, Visualization and Prediction using Data Mining
 
CRIME ANALYSIS AND PREDICTION USING MACHINE LEARNING
CRIME ANALYSIS AND PREDICTION USING MACHINE LEARNINGCRIME ANALYSIS AND PREDICTION USING MACHINE LEARNING
CRIME ANALYSIS AND PREDICTION USING MACHINE LEARNING
 
Crime Dataset Analysis for City of Chicago
Crime Dataset Analysis for City of ChicagoCrime Dataset Analysis for City of Chicago
Crime Dataset Analysis for City of Chicago
 
Predictive Modeling for Topographical Analysis of Crime Rate
Predictive Modeling for Topographical Analysis of Crime RatePredictive Modeling for Topographical Analysis of Crime Rate
Predictive Modeling for Topographical Analysis of Crime Rate
 
A Survey on Data Mining Techniques for Crime Hotspots Prediction
A Survey on Data Mining Techniques for Crime Hotspots PredictionA Survey on Data Mining Techniques for Crime Hotspots Prediction
A Survey on Data Mining Techniques for Crime Hotspots Prediction
 
PPT.pptx
PPT.pptxPPT.pptx
PPT.pptx
 
Crime Prediction and Analysis
Crime Prediction and AnalysisCrime Prediction and Analysis
Crime Prediction and Analysis
 
A predictive model for mapping crime using big data analytics
A predictive model for mapping crime using big data analyticsA predictive model for mapping crime using big data analytics
A predictive model for mapping crime using big data analytics
 
IRJET - Crime Analysis and Prediction - by using DBSCAN Algorithm
IRJET -  	  Crime Analysis and Prediction - by using DBSCAN AlgorithmIRJET -  	  Crime Analysis and Prediction - by using DBSCAN Algorithm
IRJET - Crime Analysis and Prediction - by using DBSCAN Algorithm
 
Survey on Crime Interpretation and Forecasting Using Machine Learning
Survey on Crime Interpretation and Forecasting Using Machine LearningSurvey on Crime Interpretation and Forecasting Using Machine Learning
Survey on Crime Interpretation and Forecasting Using Machine Learning
 
Life and science journal.pdf
Life and science journal.pdfLife and science journal.pdf
Life and science journal.pdf
 
Proposed Effective Solution for Cybercrime Investigation in Myanmar
Proposed Effective Solution for Cybercrime Investigation in MyanmarProposed Effective Solution for Cybercrime Investigation in Myanmar
Proposed Effective Solution for Cybercrime Investigation in Myanmar
 
Netsci
NetsciNetsci
Netsci
 
Software Defect Trend Forecasting In Open Source Projects using A Univariate ...
Software Defect Trend Forecasting In Open Source Projects using A Univariate ...Software Defect Trend Forecasting In Open Source Projects using A Univariate ...
Software Defect Trend Forecasting In Open Source Projects using A Univariate ...
 
Knowledge and Data Engineering IEEE 2015 Projects
Knowledge and Data Engineering IEEE 2015 ProjectsKnowledge and Data Engineering IEEE 2015 Projects
Knowledge and Data Engineering IEEE 2015 Projects
 
IRJET- Cyber Crime Attack Prediction
IRJET- Cyber Crime Attack PredictionIRJET- Cyber Crime Attack Prediction
IRJET- Cyber Crime Attack Prediction
 
Propose Data Mining AR-GA Model to Advance Crime analysis
Propose Data Mining AR-GA Model to Advance Crime analysisPropose Data Mining AR-GA Model to Advance Crime analysis
Propose Data Mining AR-GA Model to Advance Crime analysis
 
CRIME EXPLORATION AND FORECAST
CRIME EXPLORATION AND FORECASTCRIME EXPLORATION AND FORECAST
CRIME EXPLORATION AND FORECAST
 
Analysis of Crime Big Data using MapReduce
Analysis of Crime Big Data using MapReduceAnalysis of Crime Big Data using MapReduce
Analysis of Crime Big Data using MapReduce
 

Kyung Kim

  • 1. Reallocating Patrol Zones Based on Previous Crime Pattern Kyung Kim, Industrial and Systems Engineering Overview of Proposed Work Campus safety is one aspect that should not be overlooked when students decide which college to attend. For that reason, the Georgia Tech Police Department (GTPD) invested in new technologies and selected brave officers to ensure safety on campus. A decrease in annual crime rate at Georgia Tech for past few years shows that they have been producing fruitful results. However, our team wanted our campus to be even more harmless. With this goal in mind, in Fall 2014, we asked the GTPD what they needed help with. In our initial meeting with them, we summarized main topics we wanted to investigate on. These included resource allocation, crime clustering, and crime prediction to name a few. After a semester of work on those topics, we came up with some interesting results. We located clustered crime locations in Georgia Tech and Atlanta with help of the k-means clustering algorithm, predicted the number of crimes in the future using time series analysis, and found out that some crimes led to another crimes (e.g. stolen cars were used to commit robbery). Also, we found inconsistency of number of crimes in each zone to be the consequence of having inefficient patrol routes. However, we could fix this by incorporating results we mentioned above and suggesting a new set of patrol zones. Currently, Georgia Tech is divided into 4 zones. Among these zones, zone 2 has noticeably more crimes than any other zones. This is a natural phenomenon, as zone 2 covers more scandalous regions than others do. However, if we strategically define newly partitioned zones, then we will be able to suggest more reasonable patrol routes. We can do this by integrating our prior findings. Our data on clustering could assist us to rationally divide up the zones, and our forecasted crime locations found from time series analysis will let us check adequacy of our new zones. Therefore, for PURA and for our safety on campus, we would like to propose a new way to allocate GTPD patrol zones. Objective and Goals for the Semester Our primary goal is to make the Georgia Tech campus safer, and we can do it by increasing the police patrol efficiency by redesigning patrol zones such that the crime occurrences in each zone become approx- imately uniform. That way, the police officers have higher chance of encountering criminals and also can easily go cover for colleagues. We hope that, at the end of this semester, our results not only help the police department, but also can invoke interest from other students regarding campus safety so that more students will be encouraged to participate in research with the GTPD. We will produce a satisfying result that will make students realize what we can contribute to our campus safety. To meet our goals for this semester, we plan to complete the tasks listed below. 1. Data mining (a) Clean up data (b) Select significant variables 2. Crime pattern analysis (a) Find noticeable crime patterns (b) Look for relationships among crimes 3. Time series analysis (a) Predict future crimes i. Location ii. Time iii. Type of crime 4. Reallocate patrol zones (a) Make and solve optimization models (b) Run simulation (c) Suggest new patrol routes based on new patrol zones 1
  • 2. Methods and Techniques to be Used Our main methodology for this task will include clustering algorithms, time series analysis, and big data analysis. Some specific techniques we will apply are k-means clustering algorithms and Holt-Winters smoothing method. For our research, we will use software such as Microsoft Excel, Minitab, R-studio, Arena, Google Earth, and CrimeStat. We will use Microsoft Excel to manipulate data; Minitab and R-studio to run statistical tests; Arena to run simulation; and Google Earth and CrimeStat to run algorithms and make visual representations. Any coding will be done with Python, and it will be our main method of solving optimization models. Furthermore, we may use PredPol, a crime predicting software, if we were given a permission. We have four years of crime data from 2011 to 2014, with more than 200 variables. We will first clean up the data and remove statistically insignificant variables. Then, with the remaining data, we will apply various techniques to see what we can learn from them. As our research will be directed to observe crime patterns, areas of concentrated crimes (hot spots), and relationships among crimes, we will most likely apply algorithms to check for clustering, perform time series methods to predict crimes for 2015 and 2016, and run simulations on our model to verify appropriateness on our newly partitioned zones. Many of our optimization model will be based on mixed integer programming. As we deal with future events, we will apply robust optimization to deal with uncertainties. Our goal will be to minimize constraints from unknown possibilities and to deal with worst case to base case scenarios. When we run simulations, we will use discrete-event simulations and check for inter-arrival time, service time, and response time because this provides a way of solving complex situations, such as non-constant arriving rates [1]. We will run simulation several times, and after each run, we will make a few changes in the model and see how much improvement we can make from that. Since we will run simulation numerous times, our task will be to come up with an elegant algorithm that will closely mimic real life situations so that we could solve the problem in minimum number of iterations. Research and Background Experience Our research team has been working with Professor David Goldberg on various areas of studies before we settled on to this project. Our group has many strengths, but our greatest strength is that we are specialized. For example, we have a member who is an expert at Microsoft Excel, another member who is good at coding, and another who has a great knowledge of Statistics. I take part on working with theoretical computer science and mathematical aspects of research. Furthermore, we have diverse working experience, as we have worked for an airline industry, supply chain companies, and a technology company. Overall, we are all motivated to work hard and to produce a rewarding result. As mentioned in the overview, our research team has finished all the preliminary work [2], presented our result to GTPD, and are now prepared to begin a new project that police officers consider to be valuable for them. References [1] Zhang, Y., Brown, D.: Simulation Optimization of Police Patrol District Design Using an Adjusted Simulated Annealing Approach, Department of Systems and Information Engineering [2] Kim, K. Wang, D. Wang, Y. Bao, B. Park, S. Lee, H.: Georgia Tech and Atlanta Crime Analysis. (2015). Link: https:/drive.google.com/file/d/0B9WaQKspypsVQWJKX1Nub29sT3c/view?usp=sharing 2