Predictive Analytics for Police Forces
Transkript
Predictive Analytics for Police Forces
Predictive Analytics for Police Forces Jaap Vink Global Predictive Analytics Leader Public Sector © 2014 IBM Corporation Analytics in Law Enforcement is more than just crime prevention What if… Performance management Human resources …could see and understand all aspects impacting public safety and performance? …could attract and retain topperforming employees by understanding drivers behind employee satisfaction? Process improvement Operations management …could gain critical insights to manage and maximize performance? …could improve traffic flows and reduce emissions? Budgeting and finance Crime prevention and management …could have real-time insights into program budgets across jurisdictions to reduce spending responsibly? …could have insights to beat criminals to the scene? Programs and services …could predict current and future needs of citizens and design programs accordingly? © 2014 IBM Corporation Identifying Challenges “How can I get a holistic view of perpetrators, suspects and victims, including non-obvious relationships?” • Provide data integration capabilities with entity resolution to understand who is who and who knows who “How can I make my officers safer?” “How can I do status tracking of perpetrator / suspects ?” • • Provide statistical and historical analytic capabilities designed to define and evaluate patterns and trends Provide live situational awareness to patrol officers and detectives “How can I collaborate with and integrate information from federal, state & county agencies?” • Enable standards compliant interfaces •Enable note connectivity “How can I get predictive capability to detect, define and deter crime?” “How can I receive actionable alerts to preempt criminal activity & safety risks?” • Enable “constant analytics” to evaluate threshold situations, and provide mobile analytic capability to field police officers • Provide predictive analytics of both standard and unstructured data to deliver foresight and actionable patterns “How can I meet requirements for compliance and accountability?” • Provide supporting data and reporting capability for accountability programs and internal affairs “How can I connect the city to the public safety mission?” Enable collaboration in real time Maximize all data and sensors – even a broken street light can be the source of a mugging opportunity “How can I get flexible & robust reporting, including ad-hoc and exceptions?” • Create a robust and encompassing data architecture and infrastructure that provides ample room for expansion to meet changing needs © 2014 IBM Corporation Smarter Government A police directorate in Europe uses business intelligence to fight crime and keep roads safe 100% of crimes and traffic incidents are now reported on a centralized system >200 reports help officers and managers make better operational and strategic decisions Facilitates more efficient police work as a result of accurate, timely and location-based insight Business challenge: Years ago, the police in this region in Europe had to access 23 different systems to get the information they needed. Things got better when the region’s police directorate consolidated the information on a single platform, but the new platform did little more than spit out facts. The directorate wanted to use technology to its advantage, but to do so, it needed analytics. The smarter solution: The directorate developed a business intelligence solution that allows officers and managers to dig deeply into their data to identify trends and uncover relationships between criminals and crimes. The solution features a geographical information system (GIS) that allows users to visualize crimes and traffic incidents to identify and monitor hotspots such as accident-prone intersections, and then determine the best way to manage them. This is the country’s most comprehensive and advanced police system, and users have described it as a quantum leap in operational intelligence. BUSINESS SCENARIO: DECISIONING - FORCE DEPLOYMENT The agency uses predictive analytics to proactively deploy police forces and prevent crime Understands geographically where events are likely to happen Operational Alert Planning Shift Commander Enabling factors (Weather, Police presence, …) Risk Assessment Crash in Zone 123A If Day=Saturday And EntertainmentEvent And DayAfterPayday And DispatchZone=004 Then V_Crime=Yes (65, 0.98) Risk Assessment: area 008 Reallocation Patrol Officer If Condition=Rain And Temp>=60 And PatrolActivityPS=High Then V_Crime=No (0.92) Notices change in weather conditions Likelihood of Crime Historical crime incidents (RMS, CAD) Triggers (City events, Paydays, Time, Holiday…) Interactive Decisioning © IBM 2012 Government | Business Solution Category | Germany A State Law Enforcement Agency What if people’s safety relied on analytics? A state law enforcement agency in Germany simplifies information gathering and enables real-time access to critical decision-making information to improve public safety. The Opportunity What Makes it Smarter This German state agency manually collected information from its over 6,000 law enforcement officers, coast guard officers, criminal investigation specialists and more about every incident, every day. All of this information was spread across many systems and was too difficult to use to make immediate tactical decisions or perform longer-range planning. It is smart police procedure to record every incident attended, including all the relevant details. Up-to-date and up-to-the-minute information is critical for police work. By utilizing archived, present, and streaming data, this German police force is creating and providing a safer, and thus, smarter environment for its citizens. Combined information can now be viewed and analyzed form every angle, creating new approaches for investigation work. Pertinent information provides strategic policing, emergency response – as well as a view to potential criminal activity and scientific investigation of unsolved or present criminal events. Every law enforcement employee now has a single and easy way to input information into the department’s database – regardless of their unit, workstation or location. Real Business Results • Statistics can now be viewed in nine dimensions • Increased police response time • Increased citizen safety • Significant labor saving “The possibility of combining information and viewing it from all angles has opened up completely new approaches for investigation work.” Government | Information and Analytics | Northeast Europe Geneva Cantonal Police What if you could predict where and when crimes were most likely to happen? A police agency takes preemptive action, deploying officers where they are needed most and reducing criminal offenses, after launching a solution empowered with predictive analytics—a first of its kind in the region. The Opportunity What Makes it Smarter The mission of Geneva Cantonal Police is to ensure the safety of more than 600,000 residents. Crime in the area was rising at an alarming rate. Although police collected massive amounts of crime data from citizen calls and incident reports, the agency struggled to make meaningful use of the information. When and where were most crimes being committed? Where were future crimes most likely to occur, and by whom? Were certain types of crimes on the rise? Without concrete answers, taking preemptive action was impossible, and allocating and optimizing resources was difficult. Without greater and city-wide insight into criminal activity, the department’s efforts to anticipate, deter and stop crime were hindered. Across the globe, forward-thinking police agencies are arming themselves with sophisticated analytics to combat crime. And for good reason: It works. After implementing a predictive analytics solution, this police department gained city-wide insight into unlawful activity, enabling officers to stay one step ahead of criminals. The solution employs statistical and mathematical analytics, along with algorithmic techniques, to detect patterns and predict where and when crimes might happen. To gain contextual meaning, officers can view data interactive, geospatial maps. When thresholds for certain crimes are breached, police are alerted. For instance, the system might detect that youth offenses are more likely to be committed on Wednesday afternoons and late Saturdays and concentrated in downtown neighborhoods. The department can now take preemptive action, dispatching and allocating officers at the right place at the right time. Real Business Results • Gained the ability to better predict where and when crimes are most likely to occur • Enhanced the optimization and allocation of resources • Improved ability to determine the best course of action needed to deter crime This solution not only helps us target areas where crimes might occur, but also ensures that police officers' time is used effectively. —Didier Froidevaux, Head of Strategic Studies Department BUSINESS SCENARIO: POLICE DECISIONING - INVESTIGATIONS PD uses predictive analytics to profile crimes & criminals to improve solved crime rates and optimize resource usage Johnny is arrested for breaking into a car He is 15 years old and confesses that he wanted to belong to a group of friends Will he become a repeat offender? If YES: advise DA and later parole officer? Aspiring Repeat Offender profile … If male And age 14-16 And crime =‘car break in’ And motive =‘peer pressure’ Then repeat risk is HIGH ALERT DA … Crime Data Crime profile Team 4 A citizen reports a burglary Reports that her house was burglarized while she was talking to a representative from the city council Cluster ‘Bogus Official’ - Burglary, - Visit by city official, - Entry ‘Back door’, - Victim “Elderly’ Predictive Modeling for Crime Pattern Detection Does this crime resemble Do we have a team working on others? Is it serial? similar crimes that we can assign it to? Crime record notes and call logs Surveillance Data CS profile No Deployment A Break-in into a shop is reported The perpetrators entered by breaking a window probably between 3am and 5am. Crime was discovered at 6 pm next day Does it make sense to send out a CSI team? Is it likely that they’ll find useful evidence? An organized crime unit wants to bust a drugs ring The detectives are interested in identifying the central players within a narcotics network Who are the key persons? Who are the leaders? … If Break In And Night And report>12hrs And entry =‘broken window’ And object=‘Commercial Property’ Then probability evidence is 6% … Key Players Focus on: • Keith Patterson • Colin Wiertz • Markus Haffey © IBM 2012 Communication Data Management Dashboard Financial Data Government | Business Analytics | Southwestern Europe A European Government Agency What if you could leverage relevant information from anywhere in your organization to make problem solving exponentially more effective? A European government agency deployed a statistical analysis solution across all its departments to drive major improvements in criminal investigations and enhance its own recruitment efforts. The Opportunity What Makes It Smarter An important part of this agency’s mission is to provide support for criminal investigations. It sought to improve its effectiveness in this regard by creating a more integrated view of information – much of it highly sensitive – that was scattered across multiple departments, so that all relevant data could be brought to bear in helping resolve specific cases. The agency also wanted to improve certain aspects of its recruiting efforts. Solving criminal cases as expeditiously as possible is vital, but it is often difficult to find the “missing link” that can pull a case together. This European government agency implemented a unified analytics solution that addresses this challenge by enabling investigators and other personnel to obtain a holistic picture of crime-related data such as weapons and ballistics information, handwriting and voice identification, DNA and other information critical to solving crimes. The solution brings together millions of data points stored in multiple databases maintained by the agency’s various departments into a single source of data, upon which statistical analyses are performed to improve the agency’s effectiveness in helping to solve criminal cases. The agency is also deploying the solution to create psychological profiles, key in the staffing of specific missions, and to identify staff training needs. Real Business Results • Increased the success rate of criminal investigations • Improved the candidate selection process in staffing for critical job roles • Optimized the allocation of staff for critical tasks while reducing the risk of inappropriate behavior, absence or underperformance • Helped align organizational strategy and education programs to address key security risks as perceived by citizens “The solution has helped us sharply improve the agency’s performance in one of our principal missions – supporting criminal investigations.” German State Police Department Solving crimes made smarter, not harder The need: When solving crimes, police departments must deal with huge amounts of information. Some of this information is nothing more than tips and notices from citizens. This German State police department wanted to be able to search this type of unstructured information and find key linkages and data to support investigations. The solution: In a first-of-a-kind implementation, the department piloted the IBM an Analytics solution, developed with the help of IBM Research. The solution uses faceted, algorithm-based search capabilities to identify the specific information relevant to a case. Entity Analytics includes Relationship Resolution, which finds relevant relationships among seemingly disparate case data. What makes it smarter: Enables users to set parameters for searches, examine relationships between seemingly unrelated data, and analyze both structured and unstructured data. Allows relational searching of multiple data sources to completed electronically, instead of manually, and is anticipated to significantly reduce the person-hours required and increase the accuracy of results. Unstructured Information Management Architecture (UIMA) allows the user to leverage text analysis engines (annotators) that add semantic structure to text-based information. This allows for concept © 2014 IBM Corporation (semantic) searching. Hessian Police Criminalistic-Criminological Research Center Business Need The KKFoSt (Kriminalistisch-Kriminologische Forschungsstelle) investigates the conditions and the nature of indictable offenses and crime, with a particular focus on studying serious and repeat offenders. The aim of the KKFoSt project is to use experience in order to optimize the way in which repeat and serious offenders are dealt with and prosecuted. Solution Use advanced analysis to identify repeat offenders and those who commit serious crimes. Results • Identification of various types of repeat and serious offenders and selective categorization using cluster analysis. • Key identification of dependencies and patterns using univariate, bivariate and multivariate analytical methods. • On the basis of the MIT types identified, it was possible to derive a number of measures for investigative police work and prevention. • Prevention of criminal offenses committed by a certain type of MIT. © 2014 IBM Corporation How do you get started on the analytics strategy? 5 4 3 2 1 Focus on the biggest and highest value opportunities Start with questions, not with data Embed insights to drive actions and deliver value Keep existing capabilities while adding new ones Source: Analytics: The new path to value, a joint study by MIT Sloan Management Review and IBM Institute for Business Value. (c) Massachusetts Institute of Technology 2010. Develop an analytics plan for the future © 2014 IBM Corporation Roadmap Phase n • Flexible, phased approach • KPIs at every step • Aggressive ROI Phase n Phase n Phase n Phase n Phase 1 Time © 2014 IBM Corporation Question? Jaap Vink [email protected] © 2014 IBM Corporation