Time Magazine picked Predictive Policing model as top invention of 2011. Using Predictive policing algorithm, Cops from Santa Cruz Police Department (SCPD) are reaching the site of the crime before it takes place, thus getting ahead of the bad guys. Could a similar Predictive model have helped prevent Colorado shootout?
The recent theatre shoot out during the midnight show of “The Dark Knight Rises” on July 19th in Aurora Colorado is a far cry from your everyday crime. In the shootout, a fully armored gunman, open fired at the audience a few minutes into the movie, killing 12 and injuring 58 other movie-goers. The 24-year old alleged suspect, James E. Holmes, has apparently no history in having committed any crime before and is known by friends and neighbors as a highly intelligent man with a gift for science. So what happened? Why did this allegedly well-mannered, polite man decided to withdraw from his doctoral program last month, bought four guns all within the last couple of months and go on a rampage?
We may never get that answered as it is probably buried deep inside the folds of this suspects’ mind, but we could look at whether this incident could have been predicted and there by pre-empted. How? By using Predictive Analytics- a statistical technique by which past patterns seen in data are used to predict future events.
Last year, in July, two women were arrested after they were seen peering into cars in a parking garage, in downtown Santa Cruz. Upon arrest, one woman was found to have an outstanding warrant against her, other was carrying illegal drugs. But the presence of cops that day, in that garage was not by chance! It was part of pro-active policing done by SCPD since last year utilizing a predictive analytics system called Predictive Policing – a tool which can tell law enforcement officers where crimes are most likely to occur on that particular day.
In a recent interview with Aryng, SCPD crime analyst, Zach Friend described how SCPD was going through a difficult time last year with 20% reduction in staff and 30% increase in calls for service. With less resources available to deal with more, there were called for efficient allocation of their resources, and that they did. SCPD had a significant amount of high quality crime data; with which they contacted Dr George Mohler, Mathematics Professor at Santa Clara University. Professor Mohler along with a research team comprising of criminologist and anthropologist took all of the complex crime data and converted it into simple points that a line-level police officer could use for checks during shifts, even without understanding of algorithms and inputs.
The results: 19% reduction in burglary in last 16 months and 25+ arrests solely because of the tool. LAPD who has also deployed this same tool, has seen 25% reduction in burglaries, saving over $4 Million in costs to the community.
But burglaries are common day crimes with higher incidence vs. mass killings. Can Predictive Analytics be used in these scenarios?
Pauline Arrillaga explores how to prevent mass killings like Aurora, citing the work done by forensic clinical psychologist Dewey Cornell, whose team has developed an assessment guideline to identify threatening individuals. These guidelines are being used by most public schools to pre-empt shootouts like Columbine. Such guidelines along with other purported predictors generated with the help of crime experts (like massive purchase of guns by a person who hasn’t owned a gun before) can form the hypothesis to build a predictive model to identify threatening individuals.
Credit industries have long used an individuals’ demographic data married to purchase behavior to predict their credit worthy-ness. This same technique can be used with additional relevant data like gun purchase, crime data etc. to create a “threatening individual” score.
Whether, the current Predictive Policing model, as it stands now, can predict a crime like the Colorado shootout or not, in the near future, major police departments would probably be using a tool similar to Predictive Policing model aimed at pre-empting mass killings to create a safer and more peaceful America!