• Subscribe

How to find Banksy with maths

Speed read
  • Geographic profiling uses math to trace the source of a crime or disease.
  • Identifying addresses associated with political street artist Banksy demonstrates the reach of these algorithms.
  • Reducing a search area and search time is crucial to successfully combat infectious diseases. 

What if you could reduce a search area from 300 km2 (116 mi2) down to 16 km2 (6 mi2)? And what if your ability to find your target sooner meant saving lives? When time and money are in short supply, you’d probably opt for the smaller search area.

Researchers Steven Le Comber, Michelle Hauge, Mark Stevenson, and Kim Rossmo have recently demonstrated a geographic profiling (GP) model that could be a strong weapon against the spread of virulent diseases. To showcase the computational power of their model, they pointed their algorithms at a notorious test case: the elusive, politically charged street artist known as Banksy.

Shortening the search for a source of a disease or criminal activity is the province of GP. Employed widely in law enforcement, GP uses mathematics to connect an event (crime or infection) with its cause (suspect).

Have a malaria outbreak? Locate the mosquito-breeding site quicker and you’ll wipe it out before it reaches epidemic levels. Have a serial arsonist? The sooner he or she can be identified, the more homes and lives that can be saved.

Fingering a breeding site or arsonist are examples of GP in action, but the two cases rely on different assumptions and require different models. In the case of the arsonist, criminologists use a GP model employing a criminal geographic targeting (CMT) algorithm that assumes a buffer zone— i.e. the criminal will act in areas close (but not too close) to home. 

<strong>Marking his territory.</strong> Banksy artworks in London plotted using the Dirichlet process mixture geographic profiling model. Artwork locations are shown in red with suspect sites in blue. On the right, note the emphasis the model places on the peak in the Old Street area close to addresses Gunningham was known to frequent. Courtesy Steven Le Comber.

Nonhuman actors don’t necessarily follow this Goldilocks rule, however. There is no reason to expect a mosquito will pass up a tasty forearm just because it’s right next door to its home. For analyses that don’t assume a buffer zone, the Dirichlet process mixture (DPM) model comes in handy.

“It’s very easy to specify the probability that I commit a crime, given that we know where I live. The problem is that’s not actually what we want to know,” says Le Comber. “We want to know the probability that I live somewhere given that I commit crimes where I do.”

His team originally expected to find 10 or 15 ‘suspects,’ but quickly recognized there was only one serious ‘culprit’: Robin Gunningham, the same suspect the Daily Mail identified as Banksy back in 2008.

Using the DPM model, Le Comber’s team clustered 140 ‘crimes’ (i.e., Banksy art works) in London and Bristol, weighed them against sites associated with the chief suspect, and unleashed their algorithms to see if the suspicions were right.  

Of Banksy’s works around London, the peak of activity occurs within very close proximity to an address associated with Gunningham’s spouse and a home where he lived with a friend. In Bristol, art sites also correspond strongly with an area Gunningham was known to reside. Not conclusive, but supportive for the well-known theory that Banksy and Gunningham are the same person.

This analysis required some heavy-duty computation, far beyond the reaches of an analytical solution. “For 100 crimes, the amount of sums you have to do is around 30 million, million, million, million (3x1024) — multiplied by the number of protons in the universe,” Le Comber says with a straight face.

Finding Banksy may be trivial, but identifying the source of a disease certainly isn’t. When a pathogen or invasive species enters a new area, you’ve got to respond quickly, Le Comber notes.

<strong>Walk like an Egyptian.</strong> Map of Cairo showing locations of 139 recorded malaria cases (red circles) and eight water sources that tested positive for the malaria vectors. Geographic profiling identified these locations in minutes, whereas city officials toiled for months. Courtesy Steven Le Comber.

As an example of GP’s utility, Le Comber cites a contrast in a malaria outbreak in Cairo, Egypt between 2001-2004. Cairo officials had 139 cases of malaria strewn across 1,000 km2 (386 mi2). Narrowing this area to most likely sources still left an unwieldy 300 km2 (116 mi2) that they scoured to find potential breeding sites — standing water in old tires, ditches, water tanks, and puddles.

Using the CMT model, Le Comber reduced the search area to 16 km(6 mi2) and found in five minutes what took Cairo officials five months.

With increasing globalization and international travel, infectious diseases are on the rise. As Le Comber’s research demonstrates, GP will become a valuable tool for regional health organizations.

“What we can do with our model is to take the addresses of people with disease, and say ‘here’s where it’s coming from let’s go out and control it.’ Any thing that lets you target your interventions more efficiently has to be a good thing.”

(Banksy declined to comment for this article.)

Join the conversation

Do you have story ideas or something to contribute?
Let us know!

Copyright © 2017 Science Node ™  |  Privacy Notice  |  Sitemap

Disclaimer: While Science Node ™ does its best to provide complete and up-to-date information, it does not warrant that the information is error-free and disclaims all liability with respect to results from the use of the information.

Republish

We encourage you to republish this article online and in print, it’s free under our creative commons attribution license, but please follow some simple guidelines:
  1. You have to credit our authors.
  2. You have to credit ScienceNode.org — where possible include our logo with a link back to the original article.
  3. You can simply run the first few lines of the article and then add: “Read the full article on ScienceNode.org” containing a link back to the original article.
  4. The easiest way to get the article on your site is to embed the code below.