Skip To Content
The Mayor of London The London Assembly

Future-proofing access to crime and safety data in London

This blog reports on the recent user-led, open-source redevelopment of the SafeStats public safety data portal; with an insight into its unique functionality that has made it the leading source of multi-agency data for public safety professionals throughout London.

SafeStats, the GLA’s flagship secure portal for access to data from London’s emergency and public safety agencies, has serviced the needs of public safety professionals for over 10 years.  

It has allowed hundreds of users whose analysis has a direct impact on decision-making in service provisions, operational and strategic environments to better understand and gain insight into key safety issues.  

A key enabler in doing this has been the breadth of datasets made available in SafeStats, that may not otherwise be accessible outside of official public request pathways.  

Core to these datasets are those of London’s two police forces – the Metropolitan and City of London Police Services. Due to the agreed, ratified and regularly inspected methodology for recording and collecting, these datasets are obviously a great starting point for an initial insight into the type and volume of issues occurring. 

However, the data only includes a) those incidents reported to police that fit the crime ‘criteria’ outlined by national recording standards, and b) those that are reported to the police in the first place.  

On this second point, there is naturally a continued drive by police forces, working with partner agencies, to identify and reach out to individuals and cohorts of individuals who may require the assistance of police, but for a myriad of reasons may not wish to engage. Whilst there are existing pathways where these individuals may come to the notice of partner agencies outside the reporting of a crime to the police services (e.g. social services), how can relevant support agencies gain a comprehensive understanding of all locations and population groups that need additional resources, service provisions and targeted support? 

One option is data.  

Over the past decade, through the medium of SafeStats, the GLA has securely hosted, as well as onwardly shared data, from a range of pan-London agencies to those professionals working in community safety roles who require it. This has been through both formal data sharing arrangements with these agencies as well as the mining of available open source resources. 

These include aspects such as the locations, times, dates and types of injuries/incidents that have been recorded over the past decade by the: 

  • Metropolitan and City of London Police Services 
  • London Ambulance Service
  • British Transport Police (Trains/Tubes)
  • Transport for London (Buses)
  • Emergency Departments at London hospitals
  • London Fire Brigade
  • Royal National Lifeboat Institution (River Thames) 
  • National Drug Treatment Monitoring Service (Substance/Alcohol)

These datasets total over 25 million records. 

None of the data held in the individual datasets contains personally identifiable information, but when mapped or considered together provides analysts with a powerful and comprehensive insight into the nature of incidents occurring within a specified geographical area.  

Unfortunately, as a result of the way in which the original SafeStats system was built and the inflexibility it offered in servicing changing user needs, users were only ever able to analyse a single dataset at any one time. This made it incredibly inefficient and time-consuming to create a complete ‘London picture’ of an identified issue, as well as hindering the early identification of emerging issues across London.  

This changing user need came at an opportune time when the finance gods aligned (thank you MOPAC, TFL and LFB) to enable the complete redevelopment of the system; ensuring that SafeStats remained not only the leading source of multi-agency public safety data for London, but also sufficiently flexible to meet future requirements. 

This redevelopment programme (initially formed of a team of 1 Project Manager/Product Owner/Delivery Manager) kicked off in April 2018 with the key aims to:  

  • be user-, organisational- and market-led;
  • incorporate an open source technology stack in order to keep costs to a minimum;
  • incorporate modular architecture and functionality enabling flexibility for the future;
  • utilise agile methodologies wherever possible.

Whilst keeping the existing SafeStats system running and up to date with data, a huge amount of work was undertaken behind the scenes to create a roadmap under three initial workstreams: 

Following the first phase of development with Groundwork London’s excellent Geospatial & Data Services team, a beta release was made in October 2019 that delivered the minimum expectations of users to a) replace all back-end architecture and processing and b) to migrate all existing SafeStats functionality into a brand new web application at safestats.london.gov.uk.

Landing page

Feedback from this beta release and the remaining user needs were then incorporated into a second phase of development due for completion in Summer 2020. This phase focused primarily on the progression of SafeStats from being solely a data download tool to one with additional visualisation (through maps and charts). 

Back to that key user need of being able to consider individual themes across all the datasets and across London… 

As part of the consultation (with both users and data providers), the team were able to identify a set of ‘themes’ that existed in more than one of the SafeStats hosted datasets. 

By carrying out freetext and synonym searches on the information contained within each dataset, references to these specific themes were identified, and the record then coded appropriately for the application to recognise.  

An example being the theme “alcohol-related”: 

  • British Transport Police – where the category in the ‘offence’ field contains ‘alcohol’ or ‘drunk’ 
  • London Ambulance Service – where the category in the ‘illness’ field provided by the paramedic attending is ‘alcohol’
  • Transport for London – where the category in the incident ‘cause’ field is ‘Disturbance, alcohol related’  
  • National Drug Treatment Service – where the category in the treatment type contains ‘alcohol’. 

These individual Themes were then made available to the user in the application (below) in a way that imitated the querying of one dataset opposed to the querying of many datasets.

Selecting multiple datasets for multiple Themes
Selecting multiple areas across neighbouring boroughs
Preview of aggregated output ready for download

This process was repeated for ‘Crime Groups’ where records relevant to common crime/offence types are seen across multiple datasets.  

Clearly, there are caveats around overlap and duplication between datasets. However, through the provision of a thorough user guide that enables users to understand how these caveats are reflected in the output and how they can be accounted for, only a small number of mouse clicks are required to retrieve all the information relating to a theme. 

Users are naturally still able to ‘deep-dive’ into individual datasets more thoroughly if required by selecting the ‘Query by Data Provider’ option.

A further advantage of this improved functionality is that the SafeStats team have control over the range and content of the Themes and Crime Groups. Therefore, based on evolving user or topical need (e.g. an interest in increasing acid attacks), additional themes can be added, as and when required, to assist users in carrying out efficient and integrated analysis.  

Since its re-launch in October 2019, over 200 applications for access have been received from key public safety professionals working across the Local Authorities and Emergency Services that cover the Greater London area. There has also been a significant amount of interest from outside of London, which is testimony to the potential and increasing benefits that SafeStats can have to expand to other areas.  

There has already been a myriad of ways in which the cross-dataset Theme data has been utilised by users, from operational insight leading directly to targeted action on London’s streets, analysis of the impact of COVID-19 on different types of violence, through to trend analysis supplementing Local Authority policies and strategies.  

A recent example is Operation Lambridge which was initiated and run by the Metropolitan Police Service: 

Operation Lambridge has been the first project in London that has seen the use of ISTV data to inform targeted patrols. The project followed the principles of the “Precision Policing Model” already adopted by the NYPD but re-adapted to fit the operational and tactical policing needs of the targeted borough. The provision of timely and accurate Intelligence was one of the main pillars of the project and required a multi-agency analytical approach to ensure any available pieces of intelligence was factored in to inform adequate prioritisation of different areas and the deployment of resources where most needed.

Crime analysts are well aware of the challenges presented by the analysis of multiple datasets when investigating crime patterns: different geographies, date formats, crime categorisation and so on.. The time spent cleansing and organising the data is often more than that spent on the actual analysis. For a project requiring timely and accurate intelligence such as Op Lambridge, the ability to reduce the time spent on pre-analysis tasks has been of paramount importance.

Safestats has been an invaluable tool in reducing the data selection and data pre-processing time, allowing the ability to extract datasets about a specific type of crime – violence related incidents in the case of Op Lambridge, across multiple datasets simply through one query selection via the built-in “Crime Group” feature.

Similar to the above, the “Theme” query function has also allowed us to select weapon-enabled offences across multiple datasets without the need to research the metadata of each data provider, nor manually isolate offences/incidents that have been facilitated by the use of a weapon.

Tania Trapletti,
Victims/Violence Intelligence Analyst,
Islington Community Safety Partnership Unit

As a result of the user-led process that SafeStats has gone through, it is now set up to remain the leading source of multi-agency safety data for public safety professionals, keeping pace with future changes in user need and technology.  

For further information, please contact safestats@london.gov.uk