Want to keep learning?

This content is taken from the University of Reading & Institute for Environmental Analytics 's online course, Big Data and the Environment. Join the course to learn more.

Skip to 0 minutes and 11 seconds My name is Alan Yates. I’m a Principal Consultant at the Institute for Environmental Analytics, or IEA for short, and I’m going to introduce you to the types of projects we do, and some of the issues we find in our line of work. So let’s say you’re a decision maker with a problem. And let’s make it a big problem. It could be that you’re a supplier of fresh vegetables to retailers for a market worth several hundred million pounds per annum. But unfavourable weather conditions or extreme weather events can lead to wastage of your highly valuable produce, and you want to reduce that waste by better anticipating unfavourable weather and its impact.

Skip to 0 minutes and 50 seconds Or you’re responsible for implementing Renewable Energy projects for a remote community. You’re reliant on high cost imported fossil fuels and as a result at risk from fluctuating oil prices. You’ve been told that Renewable Energy can reduce your reliance on fossil fuels and save significant costs to your economy. But you have to quantify how much energy you can generate from your natural resources of sun, wind and wave and understand how these fluctuating sources will vary over time. ¬ .Or you are a planner for a national highways agency, whose been tasked with identifying ways to anticipate poor visibility so that you can reduce the risk of collisions.

Skip to 1 minute and 25 seconds You can think of several options: better weather forecasts, analysing traffic flows in real time or checking cctv camera footage. But how can you check out the feasibility of these ideas and build a case for further investment? Each of the decision makers in these cases have been told that there are now fantastic new sources of data available

Skip to 1 minute and 45 seconds that could help address their problems: satellite imagery; global weather and climate models or local Internet enabled monitoring devices. And at the same time there are new methods and big data techniques that could be used to turn that raw data into information. But how do you get started? How do you find out the current state of the art or best practice?

Skip to 2 minutes and 5 seconds These examples are all real cases of projects we’ve carried out at the IEA. Our role is to demonstrate the possibilities of these new data sources and techniques, often in what we call a “demonstration project”. We can help find the key relevant data, build useful models, crunch what can be very large volumes of numbers involved, and effectively visualise the results, often taking into account significant uncertainties in the data or the models. We apply our expertise in environmental data and transfer knowledge and insight to our clients. But first we need to know, what are the key issues in the client’s sector? What are they trying to achieve? What are the measures of success?

Skip to 2 minutes and 41 seconds In fact, these are key things to establish at the outset of your project, before you start to design and test solutions. Often this process is really a joint discovery with our clients, of problem and solution, and it’s the sort of challenge we love at the IEA. It’s important to understand what makes environmental analytics unique and different to all the other fields that are using data science methods today. While we use a lot of data science techniques in our work, I would say there are three things that distinguish environmental analytics from analytics in other fields, such as digital marketing or bioinformatics say.

Skip to 3 minutes and 16 seconds Firstly only in a tiny fraction of cases do we have a directly recorded high quality records of the variables that we are working with. So let’s say for example the temperature at a particular point, at a particular location in time. We probably don’t have a direct measurements, we might if we are lucky have some measurements somewhere in the locality, or we might have some measurements not at the time of interest, or we only have some low resolution estimates from satellite sources or weather models that give an indication, but they’re not going to be exact. So we spend a lot of time interpolating in time and space and trying to quantify uncertainty. And this is often the curse of environmental data.

Skip to 3 minutes and 53 seconds Secondly, in digital marketing for example, you can use what is normally well codified data to develop models to predict outputs from inputs. So let’s say a decision tree or machine learning techniques to predict if a customer will buy or not based on historic information. This is what I would call a data driven approach. In environmental analytics we’ll use these data driven modelling techniques as well, but we also have access to other physical models, that use principles say of conservation of mass or conservation of energy, to produce estimates of the things that we are interested in, say in windspeed, or pollution or flood risk.

Skip to 4 minutes and 27 seconds Because we don’t know for certain the value of environmental variables in all parts of time and space, we often end up using these models to fill the gaps. So the proficient environmental data scientist ideally know something about the available physical models Finally it seems to be the case that it’s really hard to know and find out about all the environmental data sources that may be available and relevant to your project. Even the experts can find this a challenge. There is a proliferation of publically available data sources especially in Earth Observation.

Skip to 4 minutes and 56 seconds Even though they can be open data, nominally available at no cost, you need to know what the options are, whether they are relevant, and then be able to access them and probably filter them down from what can be an unhelpfully large data source into a subset that is relevant to you and ready for use in your modelling. And ensuring you have access to all relevant data sources can often be a major time consumer.

Skip to 5 minutes and 19 seconds So that’s been a very brief look at how the IEA works with clients, and some of its unique challenges. You can read more about our projects on our website.

Business applications of environmental analytics

Listen to Alan Yates, Principal Consultant at the IEA, highlight the industries which could potentially benefit from big data analytics and the ongoing projects which the IEA team of data analysts, software developers and visualisation experts are currently working on. Alan frames the issues that businesses face when trying to get the most out of environmental analytics.

More information on the topics discussed in this video can be found on the IEA website, and Alan has also published this blog post on Data Analytics in Business.

Share this video:

This video is from the free online course:

Big Data and the Environment

University of Reading

Get a taste of this course

Find out what this course is like by previewing some of the course steps before you join: