Durham Constabulary use algorithm to predict reoffending


Durham Constabularyis using an algorithm to predict the probability that a suspect will reoffend, according to head of ICT Stuart Grainger.

The algorithm takes data such as demographic information, personal history and offending history then calculates the likelihood a suspected criminal will commit another crime, Grainger told ComputerworldUK.

This informs police choices, for example whether to release someone on bail, tag them with a GPS or keep them in custody, he explained.

It is an example of the sort of analysis the force is conducting using Microsoft Dynamics CRM, which it introduced a year ago to bring information across multiple systems into one place.

“There were immediate benefits. Officers take six and a half minutes instead of 20 minutes to enter intelligence now,” Grainger said.

However, the main benefit concerns data quality improvements, he explained. When officers enter a name into the system, it now finds the relevant individual automatically then shows any CCTV clips, 999 recordings or reports relating to that person.

The CRM also links with the automatic number plate recognition system, so officers can track vehicle movements to identify suspicious activity, he said.

“Previously, the officer wouldn’t have been able to access that. They really have a 360 degree view now. It means they are able to make decisions quickly and be more proactive,” he added.

The CRM was implemented without any formal training and has a ‘Metro style’ frontend “based on a cross between the Xbox One and traditional Metro”, according to Grainger.

“That means it follows the sort of online experience you would have at Amazon or other sites you’d use in your day-to-day life”, he said.

The system will also give officers’ access to video shot on body-worn video cameras. The force was the first to make the decision in June 2013 to issue body-worn cameras to all officers including special constables, according to Grainger.

The constabulary had to tackle difficult questions around when to turn on the cameras and whether they could infringe people’s privacy, he said.

“The technology is the easy part. It’s the engagement with the public, and making decisions around the user base, that is the stickier area.”

For the year ahead, Grainger plans to introduce “geospatial specific” alerts to officers’ devices.

“Say you walk into an area, you’ll know the priorities, what local residents are interested in, and any warrants outstanding in that area so you can go to the address and try to make an arrest,” he said.

“It’s about trying to manage resources as effectively as possible, and more proactive policing.”

Copyright © 2015 IDG Communications, Inc.

Shop Tech Products at Amazon