PredictingCrime: Identifying New Approaches and Tools
PredictingCrime: Identifying New Approaches and Tools
Theprimary role of the stakeholders in the criminal justice system is toenhance the level of security in the society. However, the continuousincrease in crime rates indicates that the current measures andresources spent in the deterrence of crime in the contemporary worldare insufficient. Federal Bureau of Investigation (2016) reported a3.9 % increase in crime rates in the U.S. in 2015 compared to 2014.The same agency predicted that the crime rate would continue growingat the same rate every year. It is estimated that about 372.6 violentcrimes and 2,487 property offenses per 100,000 citizens are committedannually (FBI, 2016). These incidents of law breaking have beentaking place in spite of the large number of people employed toenhance the homeland security. For example, it is estimated that theDepartment of Homeland Security has a total of 240,000 employees andan annual budget of over $ 41.2 billion (Kelly, 2017). However, themain challenge that security agencies and other law enforcementstakeholders face is the difficulty of predicting crimes in order toprevent them before that take place. In this paper, differentstrategies and tools used to predict crime will be discussed.
BackgroundInformation about Predictive Policing
Thestakeholders in the law enforcement sector have always wanted todevelop effective tools that can help them determine trends indifferent types of crime. This idea is based on the assumption thatthe capacity to determine the probability of certain offenses tooccur can enable the police officer to become proactive and preventthem, instead of reacting or responding to incidents that havealready taken place (Perry, McLnnis, Price, Smith & Hollywood,2013). Although there have been many attempts to invest in predictivepolicing, significant breakthroughs were made in 2008 when the LosAngeles Police Department started cooperating with the Bureau ofJustice with the objective of exploring the feasibility of theconcept of predictive policing and its capacity to deter crime (Perryetal.,2013).
Acollaboration between the government agencies attracted the attentionof other stakeholders (including researchers) who later found out in2010 that it is possible to determine the possibility of certainoffenses taking place in the same way as the scientists are able toforecast earthquakes (Perry etal.,2013). This data became a source of motivation for the stakeholdersin criminal justice sector to start developing tools and measuresthat could help them prevent incidents of law breaking. Since then,police departments in different countries (such as the U.K.) andvarious states in the U.S. (including Arizona, Washington, Illinois,South Carolina, and Tennessee) have managed to developtechnology-based tools to enable them forecast different types ofcrime and take the necessary measures to minimize the risk of theiroccurrence. The main focus of the researchers and other players inthe criminal justice system is to enhance the accuracy of thesetools.
TheApplication of Big Data and Relevant Analytics as Tools forPredicting Crimes
Thedigital age is characterized by an increase in the citizens’ andgovernments’ access to voluminous information. The big data is aconcept that was coined to describe the process in which thegovernment agencies and corporations apply technology (includingcomputer-based analytics) to take advantage of the information aswell as data that can be obtained from different offline and onlineplatforms (Perry etal.,2013). The idea of big data works by enabling the law enforcers tocollect all available information about specific categories of crimeand then using computer programs to analyze them within a short timein order to identify trends pertaining to those incidents of lawbreaking (Whetstone, Walsh, Kelling, Brodeur & Banton, 2017).These trends are then used to determine the probability of the samecrimes taking place, which enables the police officers to takeprecaution and prevent them.
Supportersof the concept of big data hold that crime has patterns that can bedetermined when a group that is large enough is observed over time.Although human conduct is hard to predict, the determination of theaverage behavior of a large population provides accurate patternsthat indicate the chances of individuals or groups of people takingcertain actions in the future (Whetstone etal.,2017). Based on this explanation, it is evident that theeffectiveness of the big data as an approach for forecasting crimesis dependent on the ability of the relevant agencies to collectvoluminous information from a large population, which is quitedifficult in the absence of technology.
Thechallenge of data gathering has been addressed by the use of moreeffective data mining tools that simply the work of the lawenforcers. Currently, there are about five algorithms and tools thatare used to facilitate mining of the big data. The first one involvesthe development of classification trees. This entails the collectionof data by classifying the dependent variables depending onmeasurements of at least one predictor variable (Hassani, Huang,Silva & Ghodsi, 2016). This helps the experts in the lawenforcement to establish nodes as well as links that enable them toapply the “if-then” rule in identifying the key sources ofrelevant data that pertains to a given type of crime that is beinganalyzed (Hassani etal.,2016). The significance of these classification trees is that theyhelp individuals and agencies determine the type of data that isrelevant out of millions of files that can be accessed offline andonline. This contributes towards the process of making it easy toapply the concept of big data to address real life issues.
Thesecond type of tool that is used in data mining is logisticregression. This is a statistical technique for the generation offormula used to forecast certain events (Hassani etal.,2016). This technique is considered to be a variant of theconventional regression, but it is extended in order to expand theconcept of classification. The role of logistic regression in theprocess of data mining is to help the expert focus on specific leadsthat will result in successful prediction of human behavior.
Third,a neural network is a type of algorithm whose development is based onthe architecture of the human brain. It was developed with numerousinput as well as the output nodes and hidden layers that facilitatethe collection and evaluation of data on human behavior (Hassani etal.,2016). Experts collect large data and give it to the input node. Thealgorithm applies the system of trial and error to adjust all weightsand help the users achieve some stopping criteria. This tool helpsthe users overcome the challenge of excess information that isavailable in the contemporary world by enabling them to sort it out.They remain with more relevant information that can allow them toforecast human behavior.
Thefourth technique is referred to as clustering and it helps users tofind groups of similar records. For example, experts are able tocalculate the distance between the point of historical data andcertain records. These records are then assigned to the nearestneighbors in a given data set (Hassani etal.,2016). The users of this technique are able to form a structuredopinion regarding the possibility of a given type of crime byexamining specific classes of attributes and grouping individualpieces of data. In addition, clustering help experts in the lawenforcement to apply the big data concept by enabling them toidentify different pieces of information since it correlates withsome examples, which makes it possible to see where ranges agree andsimilarities exists.
Fifth,the law enforcement experts can use the association technique, whichfacilitated the application of the big data technique by enabling itsusers to make simple correlations between different items. This makesit possible to identify patterns and predict the possibility of acertain crimes taking place (Hassani etal.,2016). The tendency of individuals or groups of people to adopt agiven pattern of behavior, which is identified by analyzing the bigdata, makes it possible to determine the most probable action thatshould follow a given incident. The construction of association-basedtechniques is accomplished using various tools, such as theInfoSphere Warehouse (Hassani etal.,2016). These tools provide the configuration of the flow ofinformation by examining the sources of data input, output, anddecision basis. The identification of the information flow iscritical in predicting the occurrence of crime.
TheBig Data Analytics
Thedifficulty of analyzing the large amount of information is among themost significant drawbacks that have limited the application of bigdata in the field of law enforcement. This challenge is beingaddressed through the development of the big data analytical tools.The term “big data analytics” is used to refer to thecomputer-based tools or techniques that are used to conduct theanalysis of diverse and large data sets (Ali, 2015). These tools areapplied when the size of the data sets exceeds the capacity of thetraditional techniques. Some of the key examples of tools used toanalyze the big data sets include statistical analysis, prediction,and text analytic tools that are developed by different IT companies.Their application is associated with an increase in efficiency inpolicing activities and a reduction in the cost of preventing crimes(Ali, 2015). They enable security experts to analyze an extremelyhuge data and use the outcome to identify trends, which make itpossible to forecast the probability of certain offenses takingplace. In addition, the law enforcers are able to deploy resources(such as patrol cars and police officers) in regions where crime ismore likely to occur than in other places.
TheUse of Crime Mapping Techniques to Predict Offenses
Mappingtechniques are used by the law enforcement agents to analyze andvisualize patterns of crime and identify areas where offenses arelikely to occur. This approach to predictive policing does not relyon the specific characteristics or behavior of offenders. Instead,the method is based on the likelihood of incidents of law breakingtaking place in certain geographical areas than in others (Balogun,Okeke & Chukwukere, 2014). The ability of the stakeholders in thelaw enforcement to identify the key security hotspots leads to thedevelopment of more effective policing strategies, including theallocation of more resources in the geographical locations with thehigh probability of crime. There are many mapping techniques that arebeing used to find out the security hotspots, but three of them arethe most common ones. The first one is referred to as crime analysis.It involves a statistical analysis of trends and patterns pertainingto disorder and crime (Ali, 2015). This tool is mainly used bydetectives who apply information about crime patterns in identifyingand arresting potential criminals before they engage in the lawbreaking activities.
Thesecond effective crime mapping tool is CompStat, which is the shortform of the term “Computer Statistics”. This tool was initiallydeveloped by the police department in New York, but it has beenadopted by other security agencies across the world (Meares, 2016).The system is based on the concept of target enforcement, where thelaw enforcers apply statistics to identify the specific areas whereresources are required. It is a comprehensive tool that allowssecurity officers to combine management and philosophy in policing.Most importantly, the use of comparative statistics that are analyzedfaster and accurately using computer systems leads to an effectiveprediction of crimes. This empowers the security agents to deployresources rapidly, respond in time, apply effective tactics, and makea relentless follow-up. The tool also ensures that predictivepolicing is based on accurate intelligence, instead of trial anderror techniques.
Thethird mapping technique used in predicting incidents of law breakingis the geographic information system (GIS). The GIS is a tool thathas the capacity to capture, analyze, store, present, and manage thespatial data. The GIS is also a comprehensive system that enables thesecurity officers to study different types of crime that are likelyto occur in certain geographical areas. It also enables them to findpossible causes of those crimes (Balogun, Okeke & Chukwukere,2014). This ensures that the policing measures are tailored tospecific areas, depending on the possible causes of offenses andtypes of crime, which enhances effectiveness and accuracy inmaintenance of law and order. It also contributes towards thedevelopment of proactive policing measures since the law enforcerscan foresee crimes.
SoftwareUsed To Predict Crimes
Thereare many cities and states that have developed computer-basedprograms with the objective of investing in predictive policing. Mostof the states as well as cities chose to develop their own softwarein order to address their specific security challenges.Alternatively, some jurisdictions buy programs that are available onthe shelves and customize them to suit their needs. Some of theprograms that have been proven to be effective in predicting offensesinclude the Person of Interest software, PredPol, IBM SPSS, andHitachi System.
ThePerson of Interest Software
ThePerson of Interest program is mainly used by the British policeunits. Its main purpose is to enable these law enforcers to projectwhen a given criminal is likely to strike or the probability of avulnerable person to be attacked (Edmunds, 2015). The use of the term“person of interest” means that the software was designed tomonitor and predict the behavior of individuals who have already beenidentified as potential criminals or victims. In other words, theprogram was developed with the objective of empowering policeofficers to identify individuals who are likely to commit violentcrimes at specific times and people who are at the risk of beingattacked. The software is fed with data that is sourced from the NewZealand Police and other security agencies in Europe and America. Collaboration between the British police and scholars at CardiffUniversity in the development of the program has resulted in theintegration of CCTV systems. This has allowed police officers toanalyze crowd behavior (Edmunds, 2015). The new development has madeit possible to monitor crowds and predict when fights are likely tobreak out.
Theuse of PredPol in forecasting crime
Oneof the key goals of developing a computer-based program known asPredPol that is mainly used in the U.S. was to address the issue ofshortage of resources in police departments. While discussing the aimof adopting PredPol, police chief in Los Angeles stated, “I’m notgoing to get more corps. I have to be better at using what I have”(PredPol, Inc., 2015, p. 1). The software enhances efficiency in theutilization of resources by increasing the level of accuracy of thepolicing decisions. For example, the PredPol is fed with data onrecent crimes, which is then analyzed in order to guide theadministrators in the police departments in deploying officers inareas where offences are likely to take place next (PredPol, Inc.,2015). Resources are saved by avoiding the trend in which policeofficers are deployed on the basis of trial and error or required tocarry out regular patrols, irrespective of whether offenses willoccur or not.
TheIBM SPPSS software for predictive policing
TheIMB is one of the leading technology companies in the world that haveinvested in the area of predictive policing. One of the key programsdeveloped by IBM is SPSS, which is software that allow thestakeholders in the law enforcement to collect data on incidents thatthey handle on a daily basis and analyze it to identify significanttrends (Nestler & Holmes, 2016). The IBM intended to achieveseveral objectives with its SPPS program. The main objectives includeoptimization of resources, accurate measurement of crime rates,offense reduction, improvement in the level of situational awareness,and an enhancement in budgeting as well as planning (Nestler &Holmes, 2016). One of the factors that distinguish the IBM’ssoftware from programs that are developed by most of the technologyfirms is the fact that it allows the law enforcers to use both thestructured and unstructured data to determine the probability ofcrimes taking place. The application of two types of data improvessituational understanding, which enhances the level of safety in thecommunity through effective deployment of resources.
Thissoftware was developed by Hitachi Data Systems. Although the programis relatively new in the market, it has played a critical role infacilitating the application of big data in supporting predictivepolicing. It accomplishes this purpose by providing actionable andrich insights from different data sources (Hitachi Data Systems,2016). The software makes it possible to sort out big data and selectthe most relevant pieces that can lead to effective prediction. Theprogram has been advanced over time to a level where it can beintegrated with license plate readers, gunshot sensors, andcomputer-aided dispatch in real-time (Hitachi Data Systems, 2016).This software allows experts to combine historical data withreal-time information in providing insights that increase theaccuracy of decisions. It also reduces the cost of analyzingirrelevant data. Hitachi is a comprehensive program that seeks toestablish a safer, efficient, and healthier society.
Effectivenessof Predictive Policing Tools and Strategies
Theeffectiveness of applying the modern tools and techniques to predictcrime can be measured by determining their contribution towards theachievement of the goals set by the law enforcement agencies. One ofthese goals is a reduction in the number of law breaking incidents.The use of the computer-based tools to predict crimes in 20 areas ofSanta Cruz, California resulted in a reduction in the number ofrobbery incidents by 27 % and burglaries by 11 % during the firstyear of application (Wills, 2016). This achievement is attributed tothe fact that the modern tools enable police officers to becomeproactive and take effective measures that prevent crimes fromoccurring, which goes a long way in enhancing the level of safety inthe society.
Anothermeasure of effectiveness of predictive policing is a reduction in thecost of maintaining law as well as order in the society. The tools aswell as techniques considered in this paper contribute towards adecrease in the overall cost of policing by enhancing efficiency inthe deployment of resources. For an instant, the use of tools toidentify areas where offenses are likely to occur at specific timehelp administrators in the police departments to deploy officers inhigh risk areas only (Wills, 2016). This enhances the level ofefficiency and quality of decisions made by leaders.
TheFuture of Predictive Policing
Althoughthe effectiveness of the tools used to forecast crime has beenquestioned by some stakeholders, many law enforcement stakeholdersare convinced that they will make police departments more effectivein the future. According to Wills (2016) there are four majorfactors that will advance the field of predictive policing in thefuture. The first one is the development of a plethora ofinterconnected systems that will be comprised of internet-enabledconnections, social media, body cameras, sensor networks, and CCTs.These connections will enable the law enforcers to apply bothhistorical and real-time data in predicting offenses. The secondfactor is the development of stronger IT infrastructure. Moderndatabases will be able to store large data, which will increase theutility of tools used to predict offenses. Third, artificialintelligence and the application of statistical machine learning willmake predictions more accurate (Wills, 2016). These machines willconvert raw data into actionable insights. Lastly, the increase inthe availability of smart devices (such mobile phones) will enhancesituational awareness since it will be possible to equip lawenforcers who manage crimes from offices and in the field.
Theapplication of modern tools as well as techniques to predict crimesenhances the level of safety in the society by reducing the number ofincidents of law breaking. An investment in predictive policingincreases the efficiency of the law enforcers by helping them managethe available resources and deploy them in areas where crimes aremore likely to take place. In addition, effective tools enhanceaccuracy as well as the quality of decisions made by leaders in thepolice department. The notion of predictive policing is premised onthe fact that human behavior follows certain patterns that can bestudied by analyzing a large data. This has increased the relevanceof the concept of the big data in the law enforcement sector. Themodern technology has also contributed towards the widespread use ofthe big data in predictive policing by making it possible to analyzea lot of information within a short time. The software andcomputer-based programs that have been developed by differenttechnology companies have also allowed the law enforcers to applystructured as well as unstructured data in predicting offenses, whichhas enhanced the level of efficiency.
Ali,W. (2015). Big data-driven smart policing: Big data-based patrol cardispatching. Journalof Geotechnical and Transportation Engineering,1 (2), 49-53.
Balogun,T., Okeke, H. & Chukwukere, C. (2014). Crime mapping in Nigeriausing GIS. Journalof Geographic Information Systems,6, 453-466.
Edmunds,D. (2015). Minority report: British police plan to use predictivecrime software. Breitbart.Retrieved March 27, 2017, fromhttp://www.breitbart.com/london/2015/06/07/minority-report-british-police-plan-to-use-predictive-crime-software/
FederalBureau of Investigation (2016). FBI releases 2015 crime statistics.FBI.Retrieved March 27, 2017, fromhttps://www.fbi.gov/news/pressrel/press-releases/fbi-releases-2015-crime-statistics
Hassani,H., Huang, X., Silva, E. & Ghodsi, M. (2016). A review of datamining applications in crime. StatisticalAnalysis and Data Mining,9 (3), 1-16.
HitachiData Systems (2016). Hitachi data systems unveil new advancements inpredictive policing to support safer, smarter societies. Hitachi.Retrieved March 27, 2017, fromhttps://www.hds.com/en-us/news-insights/press-releases/2015/gl150928.html
Kelly,F. (2017). Secretary of Homeland Security. U.S.Department of Homeland Security.Retrieved March 27, 2017, from https://www.dhs.gov/secretary
Meares,L. (2016). Policing should move beyond crime reduction. WashingtonPost.Retrieved March 27, 2017, fromhttps://www.washingtonpost.com/opinions/policing-should-move-beyond-crime-reduction/2016/07/20/f561df64-4cf3-11e6-aa14-e0c1087f7583_story.html?utm_term=.7258cf38b0c3
Nestler,G. & Holmes, N. (2016). Technology solutions for crimeprevention. IBM.Retrieved March 27, 2017, fromhttp://www-935.ibm.com/industries/government/solutions/crime-prediction-prevention.html
Perry,L., McLnnis, B., Price, C., Smith, C. & Hollywood, S. (2013). Therole of crime forecasting in law enforcement operations.Santa Monica: Rand Corporation.
PredPol,Inc. (2015). PredPol:The predictive policing company.Santa Cruz, CA: PredPol.
Wills,T. (2016). Anticipating criminal behavior: Data science and thefuture of predictive policing. IBM.Retrieved March27, 2017, fromhttp://www.ibmbigdatahub.com/blog/anticipating-criminal-behavior-data-science-and-future-predictive-policing
Whetstone,T., Walsh, W., Kelling, L., Brodeur, J. & Banton, M. (2017).Police. EncyclopediaBritannica.Retrieved March 327, 2017, fromhttps://www.britannica.com/topic/police