Data changemakers: Mark Brennan, Lead Data Scientist, Camden Council
In this interview Mark Brennan, Lead Data Scientist at Camden Council, discusses the data-related challenges facing the public sector and how MDM can help provide a better service to citizens.
Guy: Could start by talking to me a little bit about your own background and how you’ve ended up where you are today?
Mark: I did Geography at university back in the 1990s, so I didn’t originally have a computing background. After university I joined the British Rail computing scheme, so I really fell into data by accident, but I’ve been working in the field ever since. The British Rail scheme was great. I got into data, I started working with SQL, understanding the data and the underlying applications as well as how data and applications fitted together.
Guy: How did you find the move from the private to public sector? What are the key differences?
Mark: Moving from British Rail Computing (which became part of Atos) to Camden Council was a big change both in terms of sector and context of work as well as the associated data. In British Rail, I was primarily dealing with data around railway planning and operations with very few people related data sets. At Camden it’s totally different and I have primarily been working with resident focused data sets. We deal with some very, very sensitive data. We’ve got social services data, for example, which is extremely sensitive and if it wasn’t handled correctly the implications in terms of loss of reputation to the organisation and damage to the individuals concerned are potentially extremely serious.
At Camden I was working with sensitive data from the outset. I started working on Contact Point, a government plan for a national children’s database which has since been scrapped, but that gave me some exposure to data matching and issues of data quality in terms of how you can link records together from different business areas to get that single view of a child. Projects like that sparked the interest in the Council that eventually led the way to us implementing a single view of citizen master data management solution.
Guy: Can you tell me a little bit about your current role?
Mark: I’ve just started a new role as lead data scientist and will be leading a team of data engineers. I’ve been the technical lead for the Camden master data management solution. I also do work around enabling data quality, producing business intelligence reports for different business areas to identify problems with their data, and I work on some specific applications like upgrading our social care system. We have just set up a new Data and Analytics team here at Camden and it is exciting to be working at the heart of that.
Guy: What is it about this role that appeals to you?
Mark: I suppose there must be just something in the way my brain works! I like to see records consistent. It annoys me in my personal life if companies have got incorrect information. I’ve also got a natural curiosity so when I see an error in data I immediately start thinking “How did that happen?” and want to solve the problem.
I think it’s very important that we treat the data we hold about our citizens correctly, ensuring that it’s accurate and stored appropriately, not asking for data that we don’t need and so on. Taking this approach will enable us to perform our duties better than if records are all over the place.
I’ve always enjoyed working with things like SQL and trying to get real insights out of data, using tools to intelligently unlock the stories within the data. I love the techie side but it’s also about seeing the stories behind the data and the implications of what these records mean and what could go wrong if you don’t treat your data correctly.
I think it’s good to be slightly scared of the data, in the sense that you need to realise it’s not just a SQL table, but the information in this table contains very sensitive personal information, which, if compromised, could have a detrimental effect on somebody’s life. Or, if it’s not joined up correctly, it’ll mean caseworkers miss vital information.
I would say that certainly in terms of the sensitivity around sharing data that we’re probably ahead of the curve in the public sector. There’s got to be a public purpose to what we’re doing with the data. In the private sector you generally have a purely optional transactional relationship with organisations, whereas we’ve got a statutory duty to provide certain services so the consent model is different.
Guy: Are there any other specific data-related challenges in the public sector?
Mark: Yes, I think we’ve always got to be very careful to ensure that we’re using data in a lawful way, that we’re following the GDPR, and also that we’re transparent with the public, with the citizens about how we’re using the data. There’s been a lot of scare stories around data in the media, and certain organisations who maybe have not treated it with the respect it’s due and so there could be a fear among the public about how data is handled.
Citizens also expect that their council will provide joined-up services and that they will not have to tell the same story multiple times to different parts of the organisation. They’re used to dealing with banks and retail organisations online. There’s an expectation that they’ll be able to do the same with public services and not to have to deal with lots of different outlets into the one organisation.
This is especially important in areas like social care or supporting vulnerable families, where we don’t want to have to make people tell the same story to different professionals again and again. It is important to ensure that data is an enabler and not a blocker.
Guy: Bringing together data from numerous different systems to facilitate that single view must be a complex challenge.
Mark: Yes. We’ve got a wide range of service areas within the council from issuing parking permits to libraries to children’s social services, it covers an incredible range. If you compare this with the private sector, organisations there tend to be much more focused. It would be unusual for a private sector organisation to have to provide such a range of essentially unrelated services.
In a way, a council is like having 20 or 30 different companies under one roof, and the wide range of functionality and the different service requirements means you’ve inevitably got large number of IT systems which does result in data about a given person being on multiple different systems. Whilst we try and streamlined number of applications to keep them as low as possible, you’re never going to have an IT system which sorts out issuing library tickets and also does social care case management so there’s always going to be different IT systems.
At Camden we’ve addressed this issue through the development of our master data management solution, the Camden Residents’ Index. It provides us with that joined up view and links the different systems together so we know that a person we can identify on the library system is the same person on a housing system.
It has been a challenge to get data out of all these different systems because every system stores people’s information in a different way, for example using a different address or date of birth format, so to bring all that together was both a technical and a business challenge.
Luckily we have lots of people within the organisation who have a good understanding of the business systems and how things fit together within those systems, so we could draw upon their expertise to get the data out of the respective systems and create that single view. That in-house knowledge has been invaluable, as has the fact that most of our data was stored on premise so we were able to access it relatively easily.
Guy: So the people side of things is at least as important as the technology?
Mark: Yes, the knowledge of the people who actually use the systems is invaluable in a project like this. Applications are not always used in exactly the way that the manual would suggest. I remember an example of a system that was used for managing training courses for child-minders. The service wanted to record the dietary requirements of course attendees but there was no field within the application in which to store that information so one of the phone number fields was used instead.
When we did an extract from the system we found that some of the records had phone numbers in that field and some of them said ‘no mushrooms’ or similar. You need staff with the understanding of how your systems are used in reality to be able to make sense of things like that!
Guy: Can you talk a little bit more about the single view project. The citizen index that you mentioned?
Mark: It was both technically challenging project and a business challenging project. From the technical side, we had to learn how the technology worked which meant learning all about the ins and outs of the probabilistic matching engine, how data is weighted and how data is linked together, and how the algorithm prevents false positive matches such as twins being linked together by mistake. There was a lot of quite exciting technical work to be done.
From the business perspective, we’ve also worked closely with our legal team, people from data governance, and other key business stakeholders. It’s important to involve all these different teams because a project like this does provoke quite a lot of internal discussion.
You’ve got to have some internal honesty around the quality of your different data sources. When you implement something like this, you do learn lessons about your data and where there are gaps in the data quality and where the system can help you address them. For example, some of our housing records only had an initial rather than a first name but using the IBM technology we were able to identify what the correct first name should be from information held elsewhere in the organisation.
The technology is very good at identifying duplicates within individual business systems. That can sound like quite a dry area but the reality is that these things really matter. Imagine you’re looking at a case for a particular social care client and there’s a duplicate on the system. Maybe half the case notes are stored against one instance of that person and half of them against the other instance. Then you no longer have the full picture. It’s not just a technical problem but one that has real outcomes for the person involved. Another example might be if you have an incorrect telephone number on your system. What happens if you then need to contact that person in an emergency?
We also need to step back and think about how these data errors occurred. Are there problems with the design of some of our systems? Are they forcing a mandatory value? We found flaws in some of the way our addresses are selected in our IT systems, which we’ve been able to address with the system suppliers.
Guy: Where are you up to now with that project? What does the future look like?
Mark: The Camden Residents’ Index has been live for several years now. We’ve been adding new data sources to get a broader reach of citizen data. We’re now moving more towards a focus on data governance. To look to see about the lineage of data, to find what different attributes that actually mean. Some things which might sound really straightforward like ‘what is a child?’ can actually be quite complex. Most people would say a child was somebody under 18 but in certain services the definition will go beyond the age of 18 so it’s important to get those definitions correct.
There’s also further tuning work we can do with the algorithm to ensure better matches. It’s a bit like painting the Forth Road Bridge – you can’t just leave it unattended for too long. Our data changes all the time so we need to retune the system and make sure we’re still getting the best possible match rate.
The Camden Residents’ Index is one of the key building blocks in our wider data strategy, helping us to ensure we make best of the data we have as an organisation for the benefit of our residents.
Guy: Are there any particular skills or qualifications that you picked up along the way that you think are very helpful to you now?
Mark: The key thing I’ve learned is not to just look at things from the technical side. Much as I love the techie stuff, it’s also vital to consider the implications of what the data really means, so the ability to translate the data into real stories is really important.
I look for people who have a technical aptitude, and who are flexible and willing to learn new things and learn from others. Things change very quickly so you need to be constantly learning. I don’t particularly look for specific technical skills, but rather a general technical aptitude combined with a proactive attitude and willingness to learn. With any solution, you need to be able to relate it back to what the data represents, to work on technically challenging projects but not to have a technical tunnel vision. That’s the key thing.
Guy: Then just to wrap up, do you have any hobbies or interests outside of work?
Mark: My main hobby is running and I’m completely addicted to it. I started doing marathons in last two years and I’ve done four marathons already and I can’t shake off the habit, so I have signed up for the Reykjavik marathon in August.