Thinking Tech

Police computer predicts crimes, but is it accurate?

Posting in Cities

Predictive policing software can effectively forecast when and where a criminal might strike.

An experimental computer program, designed to predict when and where criminals will likely strike, may provide police officers with an invaluable tool as budget-strapped agencies struggle to grapple with rising crime.

Last month, the Santa Cruz Police Department began field testing the software and have recently reported that the predictions have enabled officers to pre-empt several transgressions and make five arrests, according to a report in the New York Times.

This unique approach to crime-fighting, known as predictive policing, is being trialed in several major cities like Chicago where law enforcement is turning to analytics to determine how best to ensure public safety in the face of drastic cuts in personnel and other resources.

“We’re facing a situation where we have 30 percent more calls for service but 20 percent less staff than in the year 2000, and that is going to continue to be our reality,” SCPD's crime analyst Zach Friend told the New York Times. “So we have to deploy our resources in a more effective way, and we thought this model would help.”

The pilot system being tested in Santa Cruz, considered one of the most advanced, is built on prediction models similar to ones being used to forecast post-earthquake aftershocks. To project where certain crimes are likely to occur at a certain time of the day, it crunches roughly eight years of relevant data in order to detect noticeable patterns. The results are narrowed down to a list of top 10 suggested crime "hotspots" that officers are instructed to patrol whenever they're not out responding to a distress call. Daily updates are fed into the database to ensure projections remain as accurate as possible.

The idea that such an advanced algorithm can be generated and relied upon to prevent and crime originated from a collaboration between two mathematicians, an anthropologist a criminologist and the Los Angeles Police Department, which provided the data researchers used to create the program.

It's still early, but both California law enforcement agencies seem pleased with what they've seen thus far. According to the Times report:

How accurate the program really is has yet to be demonstrated; its success will be evaluated after six months.

“The worst-case scenario is that it doesn’t work and we’re no worse off,” said Mr. Friend, who enlisted Dr. Mohler, a professor at Santa Clara University.

Mr. Friend said the early indications were encouraging. Burglaries were down 27 percent in July compared with July 2010, suggesting that the targeted policing may have a deterrent effect, he said.

In Los Angeles, Captain Malinowski said, the police department hopes to expand the program to include some violent crimes, like gang shootings.

In the article, Malinowski also mentioned that these methods may someday make predictive policing as reliable as weather forecasts. While I can see how something like this can be rather accurate and useful, I'm also skeptical since outlaws, too, can adapt rather quickly. And when they do, would the frequent updates keep the good guys ahead of the curve?

(via New York Times)

Related on SmartPlanet:

More innovative computing technologies:

Share this

Tuan Nguyen

Contributing Editor

Contributing Editor Tuan C. Nguyen is a freelance science journalist based in New York City. He has written for the U.S. News and World Report, Fox News, MSNBC, ABC News, AOL, Yahoo! News and LiveScience. Formerly, he was reporter and producer for the technology section of ABCNews.com. He holds degrees from the University of California Los Angeles and the City University of New York's Graduate School of Journalism. Follow him on Twitter. Disclosure