Computer modelling helping to anticipate crime hot spots: Report

Scientists claim to have developed a new cutting-edge computer model that predicts where burglaries are most likely to occur.

The model anticipates crime patterns, especially the hot spots of crime, which can allow law enforcement agencies to engage in targeted prevention activities that could disrupt the cause of crime before it happens, researchers said.

Using the model, the Indio Police Department, California US has developed interventions to address the problem, and can better anticipate hot spots of criminal activity and deploy officers accordingly.

The result saw an 8 percent decline in thefts in the first nine months of 2013, said researchers.

Parker began working with the Indio Police Department in 2010 to determine if a computer model could predict by census block group - the smallest geographic unit the Census Bureau uses - where burglaries were most likely to occur.

"Thefts overall had been rising, and I was concerned that we were on a course to exceed last year," said Indio Police Chief Richard P Twiss.

Using crime data and truancy records - truants account for a significant number of daytime burglaries - Parker discovered patterns of crime over time and space.

Most computer models account for changes over time or a variety of places, but not both.

"This is still cutting-edge and experimental," Parker said.

"Big data gives you statistical power to make these kinds of predictions. It makes it possible for us to anticipate crime patterns, especially hot spots of crime, which allows law enforcement agencies to engage in targeted prevention activities that could disrupt the cause of crime before the crime happens," said Parker.

Parker and Indio police reviewed 10 years of data and discovered that as truancy arrests shifted geographically in the city, burglaries appeared to follow one or two years later.

As the sociologist dug deeper into the data he identified individual students whom school officials had mailed more than 100 letters about their absences.

"We assumed there was a correlation between daytime burglaries and truancy.

"When you actually have the data that shows it, then you can evaluate the processes, and the breakdowns in the processes," Twiss said.