Big data: Not just numbers, it needs experts!
Ahead of this year’s Big Data and Analytics for Utilities conference in London this October, PEX Network Editor, Diana Davis had a chance to speak to George Taylor, the Head of Innovation at Thames Water, a major UK water supplier. Discussing the benefits that Thames Water are already seeing from Big Data investment, George also shared a key piece of advice for those interested in this space – it’s not enough just to have the data, you also need someone who knows how to use it…
Diana Davis: What do you see as the biggest opportunities for big data and analytics in reshaping utilities management?
George Taylor: I think what big data and analytics are doing is just helping us to make sense of a lot of the data that we have. Utilities are generally very asset-intensive organisations with lots of data that comes in, and I think big data now, analytics in particular, are just starting to help us make sense of a lot of that data, to spot and identify links that we perhaps wouldn’t have seen before between different inputs, different outputs and different activities.
It gives you the tools and techniques to look at really big sources of data, vast volumes of data and make sense of it; whereas, perhaps we might have relied on more traditional ways of doing that and not seen some of the things that we're now starting to see. So, it's definitely helping us to think differently and spot different trends, spot different links between things, which really helps us make some different decisions.
Diana Davis: And can you give me an example of something that you’ve been doing at Thames Water with regards to data?
George Taylor: Yes, there are a few things, really. We do some work with the Met Office, for example, when we're bringing weather data together with our leakage data, and clearly the two together start to give us some really powerful information about the links between weather and leakage. And then you can start to predict things like resources required for leakage detection purposes, or the amount of leakage you're likely to get, and the amount of work that we might have to do to respond to that, that sort of thing. Also, pollution and starting to understand what assets are most likely to cause us pollutions, and do work to fix those things, proactively, before it happens.
So again, it just starts to help us spot things, spot links between things that enable us to take action ahead of it actually happening. So, it just gives you a view on what's coming, a forecast of what’s coming, and enables you to do something about it proactively.
Diana Davis: It would seem to me, then, that setting up, say, proper reporting functions would be critical to being able to spot those trends and take proactive action. How do you go about identifying the critical metrics and then figuring out how best to present the information back to the business stakeholders?
George Taylor: Through analytics. The techniques start to help you understand the relationships between activities and outcomes. So you know, an outcome might be a blockage, but you also start to understand the required activity to prevent blockages, for example. Big data analytics can help you understand the things that you have to do to keep your blockage numbers down, and then that makes those sort of things more critical, and not just the blockage number, but the inputs, the required activities to keep your blockage numbers under control. In that case it would be cleaning, for example; you can start to determine how much cleaning you need to do in your sewerage network. There are lots of different things it gives you, perhaps, an idea of the sort of leading measures that you should be looking at as opposed to some of the live measures? So, you start to look from a reporting point of view at those things that you need to be doing, proactively, in order to make sure that your outcomes are achieved.
Diana Davis: What are the biggest big data challenges from your perspective, and what would be your advice for overcoming them?
George Taylor: I think the critical thing is doing it right. It sounds obvious, but I can knock up a spreadsheet and do a regression, it's all possible using Excel spreadsheets, but I don’t have the technical skills to actually understand some of the things I should be doing from a statistical point of view. So, I think the really important thing is that we use people with the right skills who understand what they're doing. And then, at the end of it, you use people that can make an expert, informed judgement on the outcome. Again, it's great using the models and all the things we've already talked about, it gives you a really good understanding of what you could be doing, but you still need to give the data a properly considered assessment - using someone who really knows what that data means in reality. There are two things, really, the skills of the people using the models and putting those (models) together. It's really critical and it's quite a change from more traditional engineering skills to more statistical-type skillset, which is really important in this area.