Predicting Crime with Math Model; NECESSITY Recognizing Patterns Is an Essential Part of Fighting Crime, Sheriff John Rutherford Says. FORMULAS They Determine Areas of Crime by Folding Social Patterns Together with Other Conditions

Article excerpt

Byline: DANA TREEN

For years Jacksonville and other cities have tried to predict where criminals will strike by studying neighborhood crime trends, using what has happened in the past to determine what might happen in the future.

Now, researchers believe they have developed a math model to help police identify and eliminate emerging crime hot spots.

"We can actually define where you get hot spots and where you won't," said Jeffrey Brantingham, a UCLA associate professor of anthropology who has been working to define crime patterns.

Jacksonville Sheriff John Rutherford said that kind of assessment is an essential part of fighting crime.

"That's where I wanted us to go in 2003, because I knew at that point we had to start using data to fight crime," he said. "And you couldn't just go in and saturate an area and expect that crime was going to go down. It doesn't work that way. We've done that before."

Other elements, from solving community problems to following criminals when they move, are critical to get a jump on crime.

"Now this provides an opportunity to possibly predict where it is going to go and if they can do that, that would be excellent," he said.

The research, categorized as predictive analysis, is an area of wide interest now, said Joe Ryan, the Jacksonville Sheriff's Office crime analysis administrator.

Ryan's unit does some of that now, collecting information including vacant housing numbers to bus stops and shopping center locations, then applying that to other parts of the city that are similar in makeup.

"It's community data, it's other types of data that we can look at in relation to crime events and come up with our own types of models," Ryan said.

SOCIAL PATTERNS, CONDITIONS

The work being done at UCLA is intriguing, but the next step will be critical, he said.

It is important that the research, which is based on data from Los Angeles and Long Beach police departments, can be replicated in other cities, Ryan said.

"The academics who wrote this need to do this in other jurisdictions to prove that this is something that will work universally," he said.

At the heart of the system is the development of mathematical formulas that fold social patterns of criminals and victims together with other conditions to show whether an area is a brew pot for crime.

By digging deeply into the data, the system improves on a long-standing approach that uses historical information about past crimes to predict where criminals are apt to hit next.

Crimes are classified into two types of hot spots Brantingham called subcritical and supercritical hot spots. …