Hotspot Crime Prediction
The School of Information Technology
Cincinnati, United States
Abstract— University campuses might seem very safe due to the friendly atmosphere of thousands of students, but they are often targets for planned crime. Many universities in the United States issue crime report on campus and in surrounding areas every year. Colleges with low amounts of reported crime are considered safer by people.
In this study, we present predictive system using crime data which is belong to University of Cincinnati Police Department (UCPD). Based on predictive system, we aim to decrease crimes and to predict before crime is not committed on campus and in surrounding areas, so that it will be provided take precaution against crimes by the police department.
Predictive system is an interactive system and provides to understand complicated data for police officers. The system is easy-to-use and always updatable. This system can be reached anytime and anywhere because of it works via the internet.
Keywords—predictive system; dashboard; mapping; query table; crime; university
Universities in the United States issue crime reports how many type of crimes occurred on campus and in surrounding areas every year. This is significant issue both universities and students because it affects the preference rates of the students to the universities. These reports show the crimes which are most likely to impact students in the United States – burglary, robbery, violence and sexual offences. Campus crime statistics are required by universities or federal law to report yearly to the U.S. government. On the other hand, campus safety information is not available for all schools – for example, schools that do not participate in federal financial aid programs are not required by U.S. law to report these data.
In the University of Cincinnati, crime on campus and around UC’s Uptown campus has steadily decreased since 2008. This is due to the fact that to the many proactive strategies implemented by the department and university to create a safe place to study, work and live. They have two significant strategies to decrease crime rates related our study. One of them is creation of a database to track crime patterns occurring on and around campus. Another is development of a Crime Calendar to anticipate potential increases in crime in order to allocate needed resources.
In this study, we present predictive system using crime data which is belong to University of Cincinnati Police Department (UCPD). Institute of Crime Science (ICS) at the University of Cincinnati developed a visualization system using these data and created a database which is updated daily.
In this study, we used this database that is between 2012 and 2017 and we developed our predictive system. Based on predictive system, we aim to decrease crimes and to predict before crime is not committed on campus and in surrounding areas, so that it will be provided take precaution against crimes by the police department. Also, we added a crime calendar to our system to observe the crime date and area, so that, police officers can observe the crime occurrence date and area. In this way the police will focus their resources on crime prone areas and increase their capacity to prevent crime.
Figure 1. UC Concentration of Student Residents Area (CSR)
Predictive system is an interactive visualization system that makes complicated data more interpretable for police officers. The system is easy-to-use and continuous updatable. Police easily can reach crime database using predictive system via internet. The system consists of four parts which are Crime Mapping, Heat Map, Crime Table and Statistical Data. In the following section, these parts will be explained how to use. Predictive system can predict and stop crime before it happens. Using only three data points- crime type, crime location and crime date- predictive system provides each police departments with customized crime predictions for the places and times that crimes are most likely to occur.
In sum, the contributions of this paper are as follows:
- We developed a predictive system to stop crime before it happens.
- The system provides each police departments with customized crime predictions for the places and times that crimes are most likely to occur
- Predictive system presents the observation that certain crime types tend to cluster in time and space.
- As new crimes come in, they are mapped against existing patterns and events in the city because of the system is updatable daily.
- Based on database in predictive system we predict when and where similar crimes related to these crimes are most likely to occur
Figure 2. Interactive Dashboard Predictive System
II. Background And Related Work
A few companies in the United States conduct joint projects with universities about predictive policing, but these studies are not enough and are limited. One of these is PredPol Company which is running their predictive policing project with Los Angeles Police Department (LAPD). This company do not use personal information about individuals or groups of individuals, eliminating any personal liberties and profiling concerns. An article named “Self-Exciting Point Process Modeling of Crime” was published by G. O. Mohler, M. B. Short, P. J. Brantingham, F. P. Schoenberg, and G. E. Tita in 2011. In this paper, researchers used PredPol’s algorithms
Their system is a software and uses crime type, crime location and crime date/time, and then the system can predict that crimes are most likely to occur. Our system is not a software, it is a dashboard and visualization and works via internet. Police officers don’t need a software to use our system. It is enough if they have internet connection. Also, our system is improvable. For example, next step, we can add new features to predictive system, such as; social network analysis which is very important that which criminals have contact with whom, so that law enforcement can predict people who is prone to crime.
III. Predictive System
A. Crime Mapping
As mentioned, predictive system consists of four sections. One of them is crime mapping. Predictive system uses geocoding method which is the processes of using an address to find geographic coordinates from spatial reference data. We used ArcGIS program to locate addresses in data, but this program matched 60% between crime address and physical address in reference data. As a remedy, we modified the reference data and created a new and cleaner method by writing a new code for ArcGIS. This code increased the accuracy and precision of ArcGIS, so that we could obtain better results.
Figure 3. UC Concentration of Student Residents Area Crime Mapping
Additional, as new crimes come in, they are mapped against existing patterns and events in the city. Based on the propagation patterns uncovered by the initial analysis of the data, we predict when and where similar crimes related to these crimes are most likely to occur
B. Heat Map
Heat map has very significant role to crime prediction for the places and times that crimes are probably to occur. There are two heat maps on the predictive system. First heat map is by week and year, second heat map is month and year.
Figure 4. Heat Map by week & year
Police department can analyze crime types, crime date and crime areas using heat map. They also can obtain very good results to decrease crimes and to predict before crime is not committed on campus and in surrounding areas. Heat map will be provided take precaution against crimes by the police department, so that heat map provides to take precaution against crimes by the police department.
Figure 5. Heat Map by month & year
When heat map is viewed, it will be seen dark red, medium red and light red on the dashboard. Dark red shows periods when crimes are intense. For instance, robbery crime 4 times, burglary crime 15 times, theft from auto 24 times, all other theft 17 times were reported according to the data on September in 2014. Therefore, predictive system shows September as dark red on the heat map.
C. Crime Table
Crime table is a query table on the predictive system. Query table consists of crime date and time, offense name, month, week and year (for heat map), student (yes, no), violent crime and crime address.
One of the predictive system’s aim is to protect UC’s students against crimes, because of this reason it is required whether occurred offense involves students or not. Police officers can see which crimes were committed by the students and can help them against crimes that would ever happen again.
Figure 6. Crime Table
Predictive system allows different types of queries. Police officers can search offense name or violent crime and they can analyze these results. Also, these queries that was made, appear as different types graphs on the dashboard. Crime table an easy way to create various queries for police officers.
D. Statistical Data
Statistical data consists of graphs which are row chart, flow chart and circle chart. Statistical data informs to police officers about violent crime, students who involved crime, crime statics by years and crime types, so that, the data has been becoming more understandable for police department.
Figure 7. Row chart and circle chart on the dashboard
As seen in the graphs above, violent crime and student involved graphs are seen. 725 out of 7543 crimes are violent crimes between 2012 and 2017 years. Also, 1957 students involved in a crime same between the same years
Figure 8. Row chart and flow chart on the dashboard
Figure 9. Sample coding: UC coordinates and data format (.csv)
In this study, we tried to introduce a predictive system using UCPD data. We think that this system will play a significant role to decrease crime. Especially, it can be predicted before crime happens by means of heat map and crime mapping. Law enforcement can take precaution against crimes with help of predictive system. The system provides each police departments with customized crime predictions for the places and times that crimes are most likely to occur. Based on database in predictive system we predict when and where similar crimes related to these crimes are probably to happen. It can be added new features to the system in future, such as; social network analysis system.
We thank Dr. Chengcheng Li for their insightful comments for this paper. We also thank ICS that shares their data with us.
 “U.S. News Publishes College Crime Statistics, Loan Default Data,” US News & World Report. [Online]. Available: https://www.usnews.com/education/blogs/college-rankings-blog/2014/09/09/us-news-publishes-college-crime-statistics-loan-default-data. [Accessed: 07-Apr-2017].
 “Crime Statistics,” University of Cincinnati. [Online]. Available: https://dispatcher.localhost:8443https://uc.edu/publicsafety/records-reports/crime-statistics. [Accessed: 07-Apr-2017].
 “Predict Crime | Predictive Policing Software | PredPol.” [Online]. Available: http://www.predpol.com/. [Accessed: 09-Apr-2017].
 “Self-Exciting Point Process Modeling of Crime – Google Scholar.” [Online]. Available: https://scholar.google.com/scholar?hl=en&q=Self-Exciting+Point+Process+Modeling+of+Crime&btnG=&as_sdt=1%2C18&as_sdtp=. [Accessed: 09-Apr-2017].
 “2014 Campus Crime Report final final 16.pdf.” [Online]. Available: https://www.uc.edu/content/dam/uc/publicsafety/docs/2014%20Campus%20Crime%20Report%20final%20final%2016.pdf. [Accessed: 09-Apr-2017].