FOI 428380

FREEDOM OF INFORMATION REQUEST REFERENCE NO: 428380

I write in connection with your request for information which was received on 22nd August 2025.  Please find below the response to your request:

I'm currently producing a scoping document/report around Predictive Policing, looking at how and if it is currently used within UK Police Forces?

I will be looking at how Police Forces currently do this (if so), do they use PND/Niche/Qlik and analytical methods e.g. crime pattern analysis, etc.

        - Would this then be more ADHOC, in response to a new crime pattern identified would further predictive work be carried out to assist with staffing particular hotspots or is it utilised for resourcing already?

If the force does not currently use Predictive Policing methods at present is there a reason why? E.g. Can be interpreted as discriminatory, resourcing, etc.

I wouldn't need specific information in relation to cases/crimes I'm just looking into whether Predictive Policing is utilised within UK Police Forces currently.

Reply

West Mercia Police holds information relating to your request. 

West Mercia utilise a variety of ad hoc manual analytical techniques to attempt to predict future events. This includes the use of crime pattern analysis, manually looking at patterns and trends in data and extrapolating potential future events but also some predictive techniques utilising more complex algorithms and some limited use of data science techniques. West Mercia currently uses predictive mapping to identify areas where serious violence frequently occurs and therefore is likely to occur again based on offence and arrest data extracted from the force's records management system. This comes under the Hot Spot Action Fund from the Home Office. Additional analytical methods are used to identify when previous crimes have been committed to predict when serious violence is likely to occur and targeting patrols at those times. 
 
Subject to ethics board approval, West Mercia Police is in the process of creating a Supervised Machine Learning model to identify individuals at risk of committing a serious violence offence. Individuals highlighted as at risk of committing a serious violence offence will be referred onto partner organisations and engaged in diversionary pathways with the aim of reducing their risk of offending. The model will make decisions based on the age of the individual, their gender, their previous offending history and whether they have been previously charged with an offence.