Predictive Analytics and Law Enforcement: Some Questions Arise

October 17, 2023

Vea4_thumb_thumb_thumb_thumb_thumb_t[2]Note: This essay is the work of a real and still-alive dinobaby. No smart software involved, just a dumb humanoid.

We wish we could prevent crime before it happens. With AI and predictive analytics it seems possible but Wired shares that “Predictive Policing Software Terrible At Predicting Crimes.” Plainfield, NJ’s police department purchased Geolitica predictive software and it was not a wise use go tax payer money. The Markup, a nonprofit investigative organization that wants technology serve the common good, reported Geolitica’s accuracy:

“We examined 23,631 predictions generated by Geolitica between February 25 and December 18, 2018, for the Plainfield Police Department (PD). Each prediction we analyzed from the company’s algorithm indicated that one type of crime was likely to occur in a location not patrolled by Plainfield PD. In the end, the success rate was less than half a percent. Fewer than 100 of the predictions lined up with a crime in the predicted category, that was also later reported to police.”

The Markup also analyzed predictions for robberies and aggravated results that would occur in Plainfield and it was 0.6%. Burglary predictions were worse at 0.1%.

The police weren’t really interested in using Geolitica either. They wanted to be accurate in predicting and reducing crime. The Plainfield, NJ hardly used the software and discontinued the program. Geolitica charged $20,500 for a year subscription then $15,5000 for year renewals. Geolitica had inconsistencies with information. Police found training and experience to be as effective as the predictions the software offered.

Geolitica will go out off business at the end of 2023. The law enforcement technology company SoundThinking hired Geolitica’s engineering team and will acquire some of their IP too. Police software companies are changing their products and services to manage police department data.

Crime data are important. Where crimes and victimization occur should be recorded and analyzed. Newark, New Jersey, used risk terrain modeling (RTM) to identify areas where aggravated assaults would occur. They used land data and found that vacant lots were large crime locations.

Predictive methods have value, but they also have application to specific use cases. Math is not the answer to some challenges.

Whitney Grace, October 17, 2023


Comments are closed.

  • Archives

  • Recent Posts

  • Meta