Offenders tend to commit crimes in their own backyards. When a house gets broken into, people often assume an offender from the “wrong side of town” committed the crime. “In reality, offenders search for victims locally, where they can fit in and avoid detection,” Brantingham says. “A person from the area knows the risks, knows the environment, and fits into that environment. Someone from across town doesn’t know what the risks are.”
Knowing these factors contribute to crime patterns, researchers embarked on a study using LAPD crime data to learn whether or not software could predict crime. UCLA mathematician George Mohler began adapting mathematical algorithms, testing his computer models on several thousand burglaries that occurred in the San Fernando Valley during 2004 and 2005. He found success when he adapted an algorithm used to predict aftershocks following an earthquake.
“He started tweaking that algorithm because it matched up nicely with the theory of repeat victimization,” Malinowski says.
The resulting software, PredPol, helps police agencies solve the technical challenge of optimally allocating resources. An agency’s own records management system (RMS) pushes information on date, time, crime type and location to the software and the software crunches the numbers to determine daily crime hot spots. “They don’t have access to our RMS and the data is encrypted when it goes out,” Friend explains. “But though we encrypt the data, it’s public information.”
A force multiplier
Those currently using the tool call it a force multiplier—a boon for cash-strapped agencies that have seen manpower dwindle in recent times.
“Over the last decade, we’ve had a 20 percent decline in officers but a 30 percent increase in calls for service,” Friend emphasizes. Knowing that getting more officers wouldn’t be a reality for quite some time, the Santa Cruz PD instead focused its energies on unearthing technologies designed to better allocate its finite resources. The 94-officer department volunteered to test Mohler’s software operationally and helped craft a tool that officers would actually use.
“If a technology or tool makes the police officers’ job more complicated, they’re not going to use it,” Brantingham says. “What we were looking for with Santa Cruz was to deliver the information in such a way that it was easy to obtain and did not distract officers from other tasks.”
The Web-based tool provides crime prediction boxes as small as 500-feet by 500-feet to watch commanders in less than 30 seconds, says Friend. At both the LAPD and Santa Cruz PD, watch commanders log into the software, much like they would log into a Gmail account, to obtain the 15 hot spots for that shift. They then provide officers with a list of these locations and ask them to go into them periodically during their shifts.
Though both locales have made a conscious decision to focus their use of the software on burglaries, stolen vehicles and property crime, it’s also possible to select different types of crime. Santa Cruz PD’s specialty units utilize the tool to sort out predictions for specific crimes. For instance, the gang unit clicks on a box for gang-related crimes and obtains a distinct list of hot spots matching that criterion.
“We provide the information to the officers and leave it up to them to do something with it,” says Malinowski, who admits that at first he fielded plenty of questions from officers who wanted to know what they were supposed to do with the information.
“We told them to go into the boxes and use their knowledge, skills and experience to determine what should be done,” he says.
For instance, if a box indicates an area ripe for auto theft, a seasoned officer will drive into the area and immediately notice factors that put victims at risk. For instance, it may be a neighborhood with many parked cars along a dark street. That officer might then run license plates or speak with homeowners about the risk posed to their vehicles.
In an area identified as being at high risk for property crime, an officer might talk to people in a highly visible way. “They may stay there for just 15 minutes to a half-hour and let people see them walking around the area,” says Malinowski. “Would-be offenders see the police activity and are deterred from committing a crime there. All we are trying to do is deny them the opportunity to commit that crime in that time and place. During our test, we probably disrupted criminal activity eight to 10 times a week.”