When Yorkshire Water opened their doors in May for a two-day #WaterData18 hackathon challenge to help detect water leak solutions. I was out of my comfort zone. However, using my interest in human behaviour, I pitched a new idea, the use of alternative measurements and trends, particularly use of insight to help create water detection solutions. Two people were interested enough in the subject to join me. Our team name became The Detectives. We had 48 hours to change the world!’.
The hackathon was all about the new #opendata being published by Yorkshire Water. I’m a Yorkshire Water customer and their vision is commitment to developing an ‘open by default’ approach to their data by 2020. Their ambition is customer-driven – they want to engage more with their customers and help them better understand their water supply. I took part as it was a great opportunity to learn about open data and its usefulness. Working in policy, part of my job involves working with data to inform decision making and options development based on evidence. This event was an ideal learning opportunity to develop my knowledge, and skills that can help me in my day job, to aid better collaboration between policy and analytical colleagues.
I’d only ever attended hackathons in Government and I was curious as to whether this hackathon was going to be the same or different.
The Human Touch
Having studied Psychology, I have a deep interest in human behaviour. After listening to the pitches, which included training loggers to identify sounds to better see and profile leaks accurately, a modernised listening stick, using regressions to predict the flow rates, prioritisation of district metered areas and use of drones and submarines to detect leaks!
I had basic knowledge of data and statistics from my MSc Social Research and Evaluation. I felt very much out of my comfort zone and did consider giving up and leaving within the first few hours. I stuck it out. Once as a team we identified what the scope was and understood the technical information, things got easier. Everyone helped each other sharing knowledge and talking to each other. We all got stuck in, concentrating, collaborating, researching, making and testing prototypes.
Patterns of Behaviour
The peaks on the chart show water consumption habits measured from the flow data i.e. 9am when people get ready to leave for work and 5pm when they get ready to come home. This is normal. When peaks appeared in the night – is this normal or is a leak detected? That’s the million-dollar question! If Yorkshire Water are over estimating leaks then they are sending out engineers when not needed. If under estimating leaks then they are missing out on leaks. Can we predict future occupation types based on water consumption? What’s that got to do with detection of leaks you might be wondering?
Methods
We felt the Yorkshire Water leakage calculation methods were outdated and a new modern approach was required. How can we distinguish between leaks and false alarms? We need to use futuristic methods to predict behaviour which would add more value to existing data and identify trends and insights to understand customer behaviour better.
As a team we conducted further interviews with the Yorkshire Water team to understand leakages, iterating the prototype based on new information, so that it was all joined up. What’s an actual leak and what’s a false alarm? Red herring? Confused? The leakage calculation comprised of water flow Information on non-household, house hold and suspected leakages were all aggregated together as one number. What did this actually mean? Did people understand this? What happened to transparency and openness? We felt further exploration of the problem and methods was needed to plug any gaps and spoke to other teams for inspiration and insights. The use of sketches was helpful to make associations and connections so that everything was all joined up and really helped us get closer . It helped to understand things better making the task easier including defining what was in or out of scope. We felt the insights gained from our proposed new methods calculation can ensure resources are targeted to the correct location and there is less waste as leaks are #RadicallyOpen
Customer Profiling
We proposed a customer trial on night shift workers would help us get closer to a solution. The majority of the population work from 9am to 5pm and characteristics of this group are already known, from the water flow data. By focusing on the unknowns, we can gather insight and intelligence on water consumption habits.
Prototyping
Our team got expert help from Dan Billingsley from the Open Data Institute (ODI) team, who was great to work with, to build the ‘Water Detective’.It was a shame we were racing against the clock so didn’t get a chance to get feedback from the public. The combination of data from the Water Detective and the water flow logger would generate a customer DNA profile. Information about water consumption at specific times and data from the flow logger risk status rating can help to priotise resources and send out engineers and inspectors when and where required.
We presented our findings at the show and tell and presented the various sources of evidence and recommendations for future improvements.
Paul Connell founder of ODI Leeds, was a great host. I had such a brilliant time, worked with amazing and talented people who I never would have come across before. The Yorkshire Water team were brilliant, they provided support, advice and guidance when required to everyone over the two days. I’m proud of myself that I achieved something that I thought was impossible. Those that know me, say I’m a good person to have on a team, as I notice things that others don’t and see problems through to resolution, no matter how difficult.!
I would recommend attending a hackathon, you will have fun and learn at the same time in a creative way.
‘Did we change the world? Not just yet, but all teams have been invited to present their prototype, findings and ideas to the Yorkshire Water Service Delivery Director next month. I’m hopeful that some of our ideas will be implemented by Yorkshire Water – I’ll keep you posted on progress!’
The Future is ……
Salma Afzal 