Our Online Dialogue colleague and analyst Arend Zwaneveld is for Webanalists.com started a series of articles. He looks at various real-world Big Data case studies and what we can learn from them. Today his first article: Big Data example: Crime Prevention Memphis Police Dept.

Big Data example:
Crime Prevention Memphis Police Dept.

What if you could predict crimes before they happen? This seems reserved for the ‘pre-cogs’ from the movie Minority Report, but the Memphis Police Force's Big Data solution Blue C.R.U.S.H. comes a long way: by combining historical data with real-time data, the system advises the police force where its (preventive) presence is most effective. The results are impressive: serious crime decreased by 30%, violent crime decreased by 15%.

Situation

Memphis was a bad city to live in. The crime rate there was higher than would be expected based on demographics.

Task

Scientists at the University of Memphis approached the Memphis Police Department with the idea that they might be able to discover patterns in local crime if they could access department crime data, such as geographic ‘hot spots’ on the map and times when crime is most likely to flare up.

The challenge to scientists is to ‘liberate’ data trapped in silos and translate it into concrete actions, according to the DIKW model:

• Data
• Information
• Insights
• Wisdom
• Real-time action/advice

“Most officers see police reports as a black hole,” he says. “They write a report, enter it into the system, never to see or use it again. Our goal is to start using that information usefully to solve crimes.”

Action

The scientists got to work with the statistical program SPSS, which a few years later was purchased by IBM, which then renamed the program Operation Blue C.R.U.S.H. (Criminal Reduction Utilizing Statistical History).

“Of course we knew the areas with a lot of firearm-related crime, but [Blue CRUSH's] analyses helped us see the patterns of exactly where and when the incidents occurred.”

-John Williams, Crime Analyst Unit Manager at MPS

Goal: More effective efficient deployment of police force.

The following Big Data was combined in the system for this purpose:

Result

Less crime - More crooks locked up

The results obtained from this ‘Big Data Crime Fighting’ program are worthy of note:

• 30% decrease in serious crime
• 15% decrease in violent crimes
• Higher conviction rate (from 16% to 70%) due to better burden of proof (more redacted cases)

Other police departments have also since expressed interest in the project, including those of Cincinnati, Baltimore, Boston, Las Vegas

Reusable (!) predictive models for IBM: “Industry Assets”

A Big Data system like Blue CRUSH is a self-learning system that gets better and better the more data it has and more information about the effectiveness of the recommended actions (feedback) based on the data.

For IBM, therefore, Blue CRUSH is also a commercial success: the predictive models perfected in Memphis, based on statistical analysis, can be applied in other cities without too many modifications. Developer IBM can thus resell its Blue CRUSH solution to other police forces ... around the world!

With these ever-improving ‘predictive analytics models,’ IBM can make money. For this reason, they are credited to IBM's accounting balance sheet: IBM calls this in-house acquired knowledge “Industry Assets“!

Conclusion 1: with data you can catch crooks!

Conclusion 2: good Big Data models are worth their weight 🙂 in gold!

Ethical issues in big data

Taking privacy away from convicted criminals

There is no limit to the data you can string together. How nice would it be if a suspect's location were instantly visible based on his mobile provider's data? How easy would it be for the police to receive an alert based on cell phone data as soon as the members of a group of ‘loitering youths’ get close together? In addition to fingerprints and DNA samples, will judges start taking privacy rights from convicts?

System input is human work: what is the price of a human life?

All systems that are meant to save our lives ultimately make their decisions based on data provided by people is delivered. Engineers face all sorts of ethical dilemmas, often in the form of a ‘business case’ in which a human life is expressed in a monetary amount.

More background on such ethical issues in my article ‘Big Data and Ethics.

Originally posted on July 4, 2012 at webanalists.com

Resources

[1] Frans Bentlage, “Smarter Analytics Leader Benelux” at IBM
http://new.livestream.com/eventproducent/onlinetuesday

[2] Memphis Cuts Crime With Predictive Analytics
http://www.informationweek.com/news/software/bi/226100087

[3] How To Catch a Criminal With Data
http://www.theatlanticcities.com/technology/2012/03/how-catch-criminal-data/1477/

[4] Wikipedia - Data, Information, Knowledge, Wisdom hierarchy
http://en.wikipedia.org/wiki/DIKW

[5] IBM i2 COPLINK - Accelerating Law Enforcement
http://www.i2group.com/us/products/coplink-product-line

[6] Memphis Police Department's Award Winning Real Time Crime Center (RTCC)
https://kiosk.memphispolice.org/realtime/