Geographic profiling

For Mark Stevenson‘s R package, please see here.

Geographic profiling (GP) is a statistical technique developed in criminology to identify likely candidates from large lists of suspects in cases of serial crime such as murder or rape. With large lists of suspects (268,000 names in the Yorkshire Ripper investigation in the UK in the 1980s), it is difficult or impossible to investigate each name, and a prioritisation strategy is useful. GP uses the spatial locations of crime sites to make inferences about the location of the offender’s ‘anchor point’ (usually a home, but sometimes a workplace).

A geoprofile

The model depends on two key concepts: distance decay and the buffer zone. Distance decay reflects the fact that criminals are more likely to commit crimes nearer their anchor point rather than further away, since travel requires time and effort. The buffer zone describes an area surrounding the criminal’s anchor point in which he/ she is less likely to commit crimes, either because of an increased risk of recognition, or for geometric reasons (be- cause area increases with distance squared, the number of potential crime sites will also increase with distance). The model assigns a score to each point in the search area; the higher the score, the greater the probability that the offen- der’s anchor point is located there. The resulting jeopardy surface can be overlaid on a map of the search area to produce a geoprofile. Investigators then search their list of suspects in rank order according to the height of their homes (or other anchor points) on the geoprofile. Note that GP therefore describes a search strategy, rather than giving a point estimate of the offender’s home.

GP has been highly successful in criminology, and is used by law enforcement agencies around the world. The model’s performance can be described by the hit score, the percentage of the search area which must be checked before the offender is found. For an unprioritised search, this will be, on average, 50%. Typically, GP hit scores are under 5% in criminology, a 10-fold improvement.

The success of GP in criminology has led recently to its application to biology, notably animal foraging (where it can be used to find animal nests or roosts using the locations of foraging sites as input), epidemiology (identifying disease sources from the addresses of infected individuals) and invasive species biology (using current locations to identify source populations); at the same time the model has been adapted to fit a Bayesian framework. In epidemiology, GP has been used to re-analyse Snow’s classic study of the 1854 London cholera outbreak, using 321 disease sites as input to evaluate the locations of 13 neighbourhood water pumps. The Broad Street pump (the outbreak’s source) ranked first, in the top 0.2% of the geoprofile. The same study analysed cases of malaria in Cairo, Egypt, using 139 disease case locations to rank 59 mosquitogenic local water sources, seven of which tested positive for the vector Anopheles sergentii. GP ranked six of these seven sites in positions 1– 6, all in the top 2% of the geoprofile. In both analyses the method outperformed other measures of spatial central tendency.

The method has also proven useful in the study of invasive species, where it has been used to analyse historical data from the Biological Records Centre (BRC) for 53 invasive species in Great Britain, from marine inverte- brates to woody trees, and from a wide variety of habitats. For 52 of 53, GP outperformed spatial mean, spatial median and centre of minimum distance, as well as a more sophisticated single-parameter kernel density model, especially as the number of sources (or potential sources) increased.

 

Key references (many of these can be found on my Research Gate page)

Rossmo, D.K. (1999) Geographic Profiling, CRC Press

Le Comber, S.C. et al. (2006) Geographic profiling and animal foraging. J. Theor. Biol. 240, 233–240.

Raine, N.E. et al. (2009) Geographic profiling applied to testing models of bumble-bee foraging. J. R. Soc. Interface 6, 307–319.

Le Comber, S.C. et al. (2011) Geographic profiling as a novel spatial tool for targeting infectious disease control. Int. J. Health Geog. 10, 35.

Stevenson, M.D. and Le Comber, S.C. (2012) Geographic profiling as a novel spatial tool for targeting the control of invasive species. Ecography 35, 001–012.

Le Comber, S.C. and Stevenson, M.D. (2012) From Jack The Ripper To Epidemiology and ecology. Trends Ecol Evol 27, 307-308.

Verity, R., Stevenson, M.D., Rossmo, D.K., Nichols, R.A. & Le Comber, S.C. Spatial targeting of infectious disease control: identifying multiple, unknown sources. Methods in Ecology and Evolution (in press).

Comments are closed.