About 20 people attended the OR Society’s Criminal Justice Special Interest Group meeting today in Wakefield, which was hosted by West Yorkshire Police.
Muz Janoowalla talked about a pilot project Accenture had done with the Met Police to look at developing a predictive policing tool for serious gang-related crime. Although in 2013 there were only 294 shootings of which 40% were gang-related, the public perception is that this is a much bigger issue. It is one that features within the MOPAC crime strategy and there was, therefore, a need to be able to demonstrate the impact of policing through prevention, intervention and enforcement.
They took 5 years’ data to generate a list of the highest risk individuals from a set of 12000 people within the PNC records. From 15 million records there were around 4 million duplicates.
Almost half the project was spent identifying the data and cleansing it to get to the point where predictors and visualisation models could be developed. Interestingly, the project team hadn’t given much thought to how the models might be used at the start, but it became evident that the outputs would be much more useful if engaging visualisations were used.
A set of 8 predictors was developed and every individual was given a score related to their risk of offending. Visualisations included a calendar showing dates and patterns of interaction with the police, together with mapping to show where offences were committed.
Muz concluded by explaining how the approach could be used for any crime type (given sufficient data) and that the Met was now considering how this approach could be used within an overall ‘roadmap’ for predictive policing.
Gail Mawdsley from West Yorkshire Police then talked about ‘Building an understanding of the future operational characteristics of the National Police Air Service using simulation modelling’.
She had developed a Model using Witness, with Excel inputs to look at the current service where there are 23 bases, 24 helicopters and 300 operating areas over 43 police forces. One of the main aims of this exercise was to identify options for making cost savings without adversely affecting the service.
The model was used to conduct ‘what if’ analysis of a range of scenarios with different shift patterns, base numbers and locations, aircraft configurations (including fixed wing) and task types.
NPAS has used this to inform their thinking about the service and, in February, proposed a new operating model with 15 bases able to deliver the same level of service at 14% lower cost.
One of the interesting questions asked was to what extent the data and model had been used to inform public opinion because there are often howls of protest when changes such as these are proposed. The changes are often seen as ‘cuts’, but the modelling is a powerful way to demonstrate that costs can be reduced and service levels maintained (or even improved) simultaneously.
The third speaker was Johanna Leigh, a PhD student from Loughborough University, who described an approach to developing an automated police officer selection tool for incident response.
Her work is being done with Leicestershire Police where they were trying to find a better way to help dispatchers allocate officers for emergency responses. Currently, dispatchers have a lack of information required to make good decisions on officer allocation and can end up sending the nearest officer, rather than the most suitable one.
The tool she developed takes account of road type, traffic conditions (based on historical data) and police driver status (licence type). The model, developed on MatLab, considers routing as well as decision criteria for tasking, based on discrete event simulation. It has demonstrated that, for example, response times could be reduced by 13%.
Future enhancements to the model could include the use of realtime traffic data, optimisation of officer numbers and incident queuing.
The final presentation was made by Ian Newsome and Matthew Grainger from West Yorkshire Police who presented their experience of applying design principles and analysis to challenge the policing response to people in mental health crisis. This was a pilot project carried out as part of the “Design in the public sector” initiative which was supported by the Design Council.
There is growing pressure on the police to respond more effectively to dealing with incidents where people are found to have, or known to have, mental health issues. It is sometimes poorly understood by frontline officers and their responses can therefore be inappropriate. This can result in distress for the person concerned as well as being a poor use of already stretched police resources.
The overall Design Thinking approach involved 4 stages:
WYP used Design Thinking to look at this problem in a different way. The approach aimed to help visualise the process from the service user’s perspective and involved observational techniques, including video filming interviews with users.
Having seen some of the tools used “Design Thinking” elsewhere, I’m not completely convinced that it offers much that is truly different to other improvement approaches. For example, Lean emphasises “Go and See”, 6 Sigma has “Voice of the Customer” and “Critical to Quality” and there are plenty of other well-documented tools such as Customer Journey Mapping, all of which emphasise seeing value from the customer’s perspective. However, at the end of the day, the label doesn’t matter if the result is improvement and there are better outcomes for service users.
Overall, the project employed many more visual techniques than might be used in other process improvement projects and the “value add” appeared to be in the novelty of this together with the richness of the insights it generated.
This was an excellent day, with some fascinating presentations. When the slides are available, I will post links.
Our next CJ SIG meeting will take place in London on June 5th.