Customer Success

Gloucestershire Constabulary Cleans Up

Improving data quality for operational efficiency and freedom of information

In the United Kingdom there is no central crime database, any more than there is a single monolithic police force. It is up to county and metropolitan forces to take care of their own data resources, and they all face their own challenges.

Gloucestershire Constabulary, a UK police force with 1,280 officers (plus 786 civilian staff and 170 Special Constables), is responsible for policing 560,000 people in 270,000 hectares (approximately 1,050 square miles) of urban and rural countryside. Community policing is at the heart of Gloucestershire Constabulary and in today's mobile society that depends on accurate and up-to-date information. However, criminals do not respect county boundaries, so Gloucestershire Constabulary also owes a duty of care to the other 43 police forces of England and Wales, to ensure that the data in its records are accurate.

Freedom of information
Running any kind of organization with limited resources, in the public or private sector, depends on quality information, and in 2002 Gloucestershire Constabulary embarked upon what it calls its Vision 5 programme, which involves "... providing equipment, buildings and technology that promote effective and efficient services." Vision 5 could hardly be more ambitious in its aim to make the Constabulary "the most operationally effective police force in the country". The Business Change Department in Cheltenham is working on several projects to improve operations, including IT infrastructure.

Moreover, since 1 January 2005, Gloucestershire Constabulary has had new obligations to keep its data in good order under the Freedom of Information Act (FoIA).

"These are all good reasons to invest in a solution that would address our data quality problems," says Reg Barnard, who heads the Constabulary's Information Systems Development team in Cheltenham, Gloucestershire. "The trouble is that data quality is a bit like a healthy diet and regular exercise: we're all in favor of it, but we don't always find it that easy to achieve. Every organization faces data quality issues, but in the police force, we have all the usual problems such as bad spelling and mis-keying, magnified by the fact that many of the people we are dealing with are less than entirely honest. Criminals give us all kinds of false information that can't be verified on the street, such as wrong date of birth or misspelled first names."

Of course, many of these tricks have been around for years, but with electronic data systems, it should in theory be easier to cross-check records. In practice, however, it's not so easy when you have several incompatible legacy systems, each implemented to serve different purposes.

An expanding user community
What's more the user community has grown over the past ten years, small teams of highly trained inputters have been replaced with much wider access to non-specialized users, most of them minimally trained. In other words, there are ample opportunities for inaccuracies and inconsistencies to enter the system. "With a flatter management structure, multi-skilling and multiple data entry points such as the call center, we have lost the filtering system that used to keep data quality issues to a minimum," explains Barnard. For several years the Constabulary's approach has been one of "fire fighting" the problems as they arose, but with Vision 5 and FoIA this was no longer an option.

Under the FoIA, public authorities have to produce a "publication scheme", and the Association of Chief Police Officers (ACPO) set up a national team to develop a model publication scheme for the 44 police forces. It involves delivering an open and transparent service whenever possible, publishing information that the public has an interest in viewing and making specific information available on request, so long as this does not prejudice ongoing operations.

The SAS solution
SAS recommended that they deploy SAS Enterprise Data Integration Server, part of the SAS^®9 Intelligence Platform. This solution includes dedicated data quality and data integration interfaces, namely DataFlux dfPower Studio and SAS Data Integration Studio. "We wanted a solution that would not only enable us to address our immediate data quality concerns, but also lay the foundations for business intelligence," says Barnard. "After all, good policing is about quality intelligence: making the link between the crime and the offender."

That's a view echoed by Nick Churchill, who spent 31 years as a detective and is now the Constabulary's Database Administrator. In 2003 he was given the task of shifting the data from the legacy systems into a data warehouse and making it available on the intranet. He chose SAS for a pilot project that would integrate three legacy systems: Unity, which is the core intelligence system containing data on crime and custody; VPFPO, which stores data on vehicle parking and fixed penalty offences; and the Domestic Violence database, which contains details of offenders and victims. "We did not make things easy for SAS. All three systems are implemented with incompatible software: Oracle, Ingres and SQL," says Churchill.

All of the information extracted from the source systems is subjected to SAS data quality procedures to profile, cleanse and standardize the data, using match coding. Gloucestershire Constabulary found that with DataFlux dfPower Studio you can adjust the sensitivity on matching fields, such as name or street address. That way you can identify the biggest problems first and work downwards. It will quickly pick up obvious alternative spellings of the same name, such as Stephen and Steven or Uddin and Udden, but it will also pick up less obvious matches. For example, the surname Ferry might be accurately recorded on the street, but then wrongly entered as Perry back at the station. Likewise, car registration numbers might include one or two wrong digits. You would have a hard time finding such mistakes with other software, let alone manually.

Safety first
"However, there is an important caveat. One thing I dare not do is change the data in Unity, because how do I know for sure whether two very similar records relate to one person or two? It's a very risky business: if you get it wrong you could obstruct an investigation or you could make unjust assumptions about an individual," says Churchill.

Consequently Gloucestershire Constabulary is using SAS to identify possible quality issues, but feeds the questionable records back to the owners for auditing and correction. A typical problem is wrong postcodes and addresses, which a police officer on the beat has no means of validating. It is relatively easy to identify mismatches, but then the question arises, is the postcode wrong or the street address? One way of isolating such issues is to match coding between the different data sources. "Much to our surprise, a quarter of the vehicles on the VPFPO system were also on the Unity system, meaning they were used in a crime of some sort, with or without the knowledge of the owner. Yet, very often, the least reliable and up-to-date information is on Unity, our main operational database.

"SAS is very powerful when it comes to clustering the near-matches to help us correct data and bring it up to date, so it saves a huge amount of time and resources," says Churchill.

"What's more, SAS Data Integration Studio is really easy to use, considering how powerful it is. The drag-and-drop technology means you don't need any specialist programming knowledge, just some basic knowledge of the data you are interrogating. In no time you are producing reports that will amaze senior managers."

Police officers typically work in difficult circumstances where mistakes are inevitable; mistakes that ordinary database software will not identify, even when it is working properly. SAS offers the additional layer of intelligence that will now enable Gloucestershire Constabulary not only to spot individual mistakes, but also to correct procedures and identify training needs. "With poor quality data, we were wasting a lot of time trying to find out things we already knew. SAS is putting us on track to achieve the operational efficiency mandated in Vision 5 and to meet our obligations under FoIA. In practical terms, it could ensure that a police officer arrives at the right address when trying to apprehend a criminal," concludes Barnard.