Improving the location anonymisation

Hi,

We are currently working on improving the way in which locations on the crime map are anonymised and would like to hear your thoughts.

As you may be aware the crime data is currently anonymised using ‘Snap Points’, predetermined locations that adhere to several criteria (e.g. they are in the middle of the road, not within an area with less than 8 addresses; more information can be found here http://data.police.uk/about/#location-anonymisation).

Before we make any changes we’d  like to get a better understanding of what factors are important to our data users, and what changes we could make that would maximise the usefulness of the data. To kick off the conversation, there’s a few questions that we’d welcome your thoughts on:

  • Do you think it would be useful to provide exact location data for public spaces (like parks, car parks) where privacy is not a concern?
  • How important is it for a crime to snap to it’s original street? Would it be better to snap the crime to a point on a different road, but closer to the true location?
  • Would there be any negative impact of having multiple ‘snap points’ on the same street?
  • For crimes that occur near a police force boundary, is it OK to anonymise it to a close location over the force boundary, or better to always anonymise it within the original force but potentially further away?

Any opinions, ideas and suggestions would be much appreciated and will help us shape the service in the future.

Comments

Improving location anonymisation

The primary changes I would like to see to the anonymisation process are:

  • prevention of snapping across administrative boundaries,
  • provision of accurate co-ordinates for public places and
  • publication of the polygons used for allocating crimes to snap points.

Until late last year I ran a website at crimeinlondon.com that provided ward-level statistics for crime in London showing changes over time, comparison to similar wards elsewhere, neighbouring wards and so on. I closed it down after complaints from police officers that the crime counts I was publishing were not the same as those that they saw internally. This essentially made the site useless for holding local police to account, because any citizen asking their neighbourhood team about an increase in a particular type of crime would simply be told that the figures I was publishing were wrong.

What is needed is a way to make police.uk data useful for holding police to account by allowing comparison between areas and tracking over time. This could be done by guaranteeing that crime points will always be snapped within a particular level of administrative or census geography, for example always snapping locations to a point within the same ward in which the crime actually occurred. This could be done with lower error in urban areas than in rural areas, but if a distance parameter were added to the data (showing how many metres between the crime location and the snap point) then data users would be able to gauge the accuracy of the data in any given area and warn end users where appropriate. Obviously the larger the area to which the guarantee applied, the less useful it would be. I do not believe it is particularly meaningful to guarantee snapping only within such large areas as local authorities or police-force areas. Conversely the snapping distance would increase greatly if snapping was guaranteed to be within very small areas such as output areas or LSOAs, so this might not be feasible. Local authority or census wards are likely to be a reasonable compromise, with the added bonus that many police forces use ward boundaries as the boundaries of neighbourhood-team areas.

Provision of accurate co-ordinates for crimes occurring in public places would obviously be beneficial, not least because it would stop streets close to high-crime public places from appearing themselves to have lots of crime due to snapping. If this is done, a new column should be added to the data to show whether the location provided is the actual or snapped location, so that data users can present locations with confidence and end users can have confidence in the quality of the data provided by the Home Office.

One of the big problems with using the police.uk data at present is that data users are left to guess as to which area/street segment is covered by each snap point. This makes it impossible to say 'there were n burglaries on High Street' because we can't be sure if all the crimes snapped to that point actually happened on that street. If the Home Office believes the street segment is the fundamental unit at which crime data should be published, it should make sure data actually relate to street segments rather than merely appearing to do so. For the sake of transparency, the Home Office should also publish polygons showing the area covered be each snap point.

 Flag as offensive 

Improving the location anonymisation

I support the publication of crime point data.  Clearly there will be glitches (I do believe that there should be a discussion about whether for instance domestic violence is reported).  However car crimes, theft, night time economy violence and robberies should be point data - and located on streets.  People who report crime know something happened; people who live in the area may know also - unless the victim does not live in the area - but there should be clear pressure to require attention on the places that cause most harm.  This is an evidence based criminological finding and the public deserve to know where these places are and what police are doing about them.

 Flag as offensive 

Do you think it would be

Do you think it would be useful to provide exact location data for public spaces (like parks, car parks) where privacy is not a concern?
Definetley yes, you want to feel save when leaving your car somewhere.
 

 Flag as offensive 

Hello everyone, thank you

Hello everyone, thank you very much for the responses so far!

We had some great feedback both here and via email, we will have a closer look at all of them in near future and decide what can be feasibly implemented. At first sight it appears that both mapping exact locations in public spaces and having multiple points on long streets are popular ideas so we will definitely consider taking them further...

Matt, I like the idea of showing what areas snap points relate to - I will pass it on and we will see if this is something that could be made public on the website.

 Flag as offensive 

I agree with you ("One of the

I agree with you ("One of the big problems with using the police.uk data at present is that data users are left to guess as to which area/street segment is covered by each snap point. This makes it impossible to say 'there were n burglaries on High Street' because we can't be sure if all the crimes snapped to that point actually happened on that street. If the Home Office believes the street segment is the fundamental unit at which crime data should be published, it should make sure data actually relate to street segments rather than merely appearing to do so. For the sake of transparency, the Home Office should also publish polygons showing the area covered be each snap point.")

 Flag as offensive