Digital dragnet: How data became a cop's best weapon
- By Patrick Marshall
- Nov 29, 2011
Grand Junction, Colo., doesn’t usually have a lot of serious crime. So when a series of armed robberies occurred one Saturday morning last summer, followed by a burglary and then a homicide in the afternoon, police wondered whether there was a connection.
“We had very little suspect information on these crimes,” said Troy Smith, deputy chief of the Grand Junction Police Department in Colorado. But CopLink, a data-sharing and analysis tool, helped police connect the dots and identify suspects within 12 hours.
“The suspects were from California,” Smith said. “We began very quickly working with agencies in California, who have the same software system, and we were able to share information about the suspects. We were able to really lock in pretty rapidly.”
'Google for police' goes mobile
Taking criminal justice resources into the wild
Driven by shrinking budgets, and encouraged with federal grants, local law enforcement agencies have increasingly been turning to technologies originally developed for business analysis to more efficiently battle crime.
“Police departments have long had tons of data at their disposal,” Smith said. “But the data that we had for years sat either in paper form or in computers and was never looked at. The difference today is that we have really a whole new breed of employees in police organizations who are really proficient at analyzing that data using different pieces of software and beginning to put together patterns that exist in that data between jurisdictions that allow us to deploy our resources much smarter than we've ever been able to do before.”
With economic conditions shrinking available resources, Smith said, “we have to be in a position to turn what looks like data in a computer into a briefing sheet or operations plan that changes how we deploy our resources in the field.”
Vendors are aware of the need — and the opportunity — and are rushing to fill it, tailoring technologies that were originally designed for business analytics to better suit the needs of law enforcement.
“There's also demand to provide the same level of service or more with the same amount of resources or less,” said Bob Griffin, CEO of i2, the maker of CopLink. “People are looking for that competitive edge.”
Vince Talucci, principal adviser for state and local law enforcement at SAS, the North Carolina-based business analytics software company, agreed.
“The decrease in resources, combined with the onslaught of never-ending data really brings us to a tipping point for the value of analytics to a given agency,” Talucci said. “While we have solutions for investigatory matters, we are very much trying to focus on prevention of crime and disorder, but doing it in a very mindful way of the realities facing law enforcement.”
Although using computers to collect and analyze crime data is not new, the tools are quickly growing more sophisticated and effective.
“When we started this more than 10 years ago, we talked about things like information sharing and gathering as much data as you could,” Griffin said. “In those days, people would look at you like you're a green banana. Why would I want to share information? Why would I want to bring information from business licenses or hunting and fishing licenses into the policing environment? Well, there are lots of reasons for it.”
The initial focus in most jurisdictions is on integrating databases within the jurisdiction so that analytics tools can be applied to find patterns in the data and user queries can be applied across all the data. “It's easier the more data you have,” Griffin said. “It gives you more bang for the buck.”
“Currently we have data from a record management system and our [vehicle] tab system along with some intelligence data,” saID Jason Scheiss, analytical services division manager at the Durham Police Department in North Carolina.
The department uses two tools from i2 — iBase and Analyst’s Notebook — to integrate the databases and to perform queries and pattern analysis. “Yes, we’re absolutely looking to expand the data to include, perhaps, the city's utilities database, which includes individuals who have water and sewer hookups, and we could even use it to bring in the visitor logs from parks and recreation facilities. We could bring in the daily jail list. So we could say, ‘Hey, look here. All of these crimes only occur when this one guy’s not in jail.’”
License plate readers
Scheiss even suggested integrating data from automatic license plate readers on federal interstate highways.
Most vendors offer a variety of modules that allow departments and agencies to design systems to include just the data they want to track. For example, i2 offers a module that performs infrastructure monitoring and processes the data into iBase so that it can be analyzed. “If people gather around a water reservoir, for example, somebody should be aware of that,” Griffin said, “because there may be something greater and more nefarious going on than just a simple trespass.”
An analyst might then receive an alert via Analyst’s Notebook. And if the department or agency is using CopLink, the information can be sent directly to officers in the field. And, as of October 2011, CopLink is available for smart phones and tablet devices.
The state of North Carolina turned to SAS when it decided in 2008 that it needed to develop a way to integrate and analyze data across the state’s various criminal justice applications. The result was Criminal Justice Law Enforcement Automated Data Services (CJLEADS).
“We had two very tragic murders at two of our flagship universities in January and March 2008,” said Kay Meyer, CJLEADS program director. “It became apparent that the individuals involved had also had encounters with law enforcement and the courts in the weeks prior to the murders. They had probation violations and outstanding orders for arrest. But because those warrants were from other counties, that information was not easily available to the people who had interaction with them in the weeks prior to the murder, so instead of being in custody like they should've been, the second murder occurred.”
The state legislature quickly called for the development of an integrated data system. After a testing period, statewide deployment of CJLEADS was launched in January 2011.
“It gives us one place where we can log in and see everything from a person’s court records, their outstanding orders for arrest or warrants,” Meyer said. “From our Department of Corrections, we can see your incarceration information. We also have information if you’ve ever been on probation or parole. We have your parole officer’s name and other charges associated with your supervision. We also show jail data. We have our sex offender registry. We have some gang information as well as concealed handgun permits. And we have a real-time interface with DMV, so we can show driver’s license information as well as vehicle information. Very, very quickly we’re able to see all the information about this person.”
But making all the data available in one place is only half of what CJLEADS does. “Our second objective was to build a watch list capability,” Meyer said. “If you have a case that you are working on that involves two or three people, you can add them to your watch list. The system will alert you if those individuals have any kind of change in status. If they are arrested somewhere in the state, if they are released from custody, if they have an upcoming court date that has changed, you receive an e-mail alert that something has changed about that person.”
Some jurisdictions are also looking to analytic tools to help predict when and where crimes are likely to take place. “We are not predicting where the next crime is going to occur,” Talucci said. “What we are focused on is forecasting the probability of occurrence based on a bunch of different variables to get the best deployment of limited resources.”
The police department in Santa Cruz, Calif., is in the middle of a six-month test of a predictive application that does not even have a name yet. “It is a Web-based system that we use to input data into an algorithm that produces 10 hot spot maps a day,” said Zach Friend, the department’s public information officer and management analyst.
Friend said the department ran tests on historical crime data and the algorithm predicted about one-third of the crimes within a given location. “So we decided to implement the program,” he said.
“The output is pretty simple,” Friend added. The program creates maps of 10 500-square-foot locations in the city. The maps show the highest probability of two one-hour time windows during which crime might occur, as well as the probability of what the crime will occur against. “If it's a burglary, are we looking at the street for cars or are we looking up the driveways and down the alleyways to see if it's a home that is the target,” he said.
“We did our first review at the end of September,” Friend said. “We did see a strong correlation between the number of extra checks we did and a reduction in the crime types that we were targeting. In addition, there was a reduction of crime for about two-thirds of a mile around that area. We did not see displacement of crime. Just being there didn't mean it just went six blocks away.”
i2’s Griffin agrees that there is a future for predictive policing. “The whole idea behind predictive policing is to allow you to be more effective from a force deployment perspective,” Griffin said. “That is why we built our dashboard technology, which allows us to go off and look at events from a real-time perspective and send out alerts. So you can say somebody should pay attention this particular set of events because they're clustering [in] a certain location and they're clustering at a certain time.”
Smith said his department was able to use the tools in i2’s Analyst’s Notebook to end a string of auto thefts. “We were able to use a computer model to predict the next likely location where this would occur based on the previous pattern,” he said. “So we deployed some resources and some surveillance on that location, and within about 45 minutes we had two suspects in custody.”
Learning to share
Although integrating databases within a jurisdiction is generally the first step for most departments and agencies, there is a lot to be gained from sharing data among jurisdictions as well.
“Eight agencies in the state of Colorado came together and really saw the need for us to be able to access and share each other's information, so much so that those agencies subsequently purchased a statewide license for CopLink,” Smith said.
“I think it's the greatest thing that we could've done in terms of increasing officer and community safety in the state of Colorado," Smith said. "An officer in eastern Colorado can have at their disposal in the police car an offense report involving a suspect in the western part of the state right here in Grand Junction. Having the ability to pull that information up at the scene of an incident can completely change the type of interview that officer does, the way they approach a situation.”
According to Smith, 76 agencies in Colorado, representing about 85 percent of all law enforcement officers, are now using the system.
Smith also stressed the importance of being able to share data across state borders. “Our society is incredibly transient,” he said. “People can be in Boston this afternoon and tomorrow morning they're in L.A. And it's not just businessmen and entrepreneurs who are traveling across our country. It is also people who are engaged in criminal enterprises. So it is incredibly important for us to be able to share this information. It's still a work in progress for us.”