Can data science help create a sustainable future?

Codepan has always been future-focused. Since its founding, we’ve striven to build technologies which contribute to the advancement of society. In line with this perspective, this autumn Codepan decided to realign its company purpose to what we believe is the most pressing issue of our times, to drive the industrial sustainability revolution using data science.

So what does sustainability mean?

It is rightly an ongoing discussion, but one main authority on the matter is the United Nations. In 2015, the UN agreed upon 17 sustainable development goals (SDGs) which organisations and countries can align themselves towards to create a better and more sustainable society [1]. Sustainability is meeting the needs of the present without compromising the ability of future generations to meet their own.

How does data science fit in?

One could argue that “sustainability data science” is where the needs of these sustainable development goals can be met with data science. We’ll look at some use-cases later in the post, but for now here are some important questions that excite us, and that data science can ask in this field[2]:

  • How can fault detectors be applied to agricultural water conservation?
  • In what ways can building energy efficiency be automated?
  • How can cell-phone data inform transport planning?
  • How do we optimize data centres to reduce wasted heat and energy consumption?
  • In what ways can we monitor and improve solar panel energy generation?

So we just make our machines more efficient?
It was first observed in the dawn of the industrial revolution, that improvements in efficiency can increase demand. Jevon’s paradox [3] highlights that rather than mitigating them, increases in efficiency could ultimately lead to an increase in greenhouse gas emissions. Green markets as engines for a sustainable transition can, therefore, be challenging, as the sustainability impact of a product or service isn’t necessarily assessed before it is sold or traded. Whilst one product may result in a positive economic outcome, the negative externalities on society and the environment may not be duly considered. We want to get this right, and this requires repeat visits to potentially difficult reflections.

The Cambrian-like explosion of data we’re currently seeing provides a huge potential for novel products and business insights. To make the best of this great opportunity, we must realign these additional environmental and societal costs as first-class costs alongside economic costs.

“Sustainability is meeting the needs of the present without compromising the ability of future generations to meet their own.”

Sustainability and the climate crisis

It is necessary to describe the current state of the environment in terms which reflect the severity of the problem at hand [4, 5, 6]. The 13th SDG acknowledges we must “take urgent action to combat climate change and its impacts’’. Unleashing the extensive potential of data science to the mitigation of greenhouse gases (GHG) and adaptation to a changing climate is key to this. Spaces where domain experts meet software engineers and data scientists are the spaces where novel solutions to reducing these negative externalities can evolve.

Our niche

Codepan has always been on the cutting edge of technology. Whilst working on full-stack development projects, including clients such as Fleetboard and Infarm, our main focus has been anomaly detection, from which we generated, an automated detection tool.

In the context of sustainability, time-series analysis touches multiple sectors [7]:

  • demand forecasting in electricity grids
  • optimizing and modelling public transportation networks
  • automated monitoring & predictive maintenance of energy pipelines
  • early-warning systems for natural disasters
  • food production monitoring

We’ve build machine learning solutions for a myriad of challenges in the industrial sectors, always finding the right tool for the job. We will go deeper into some use-cases later, with a future blog post of other projects we’ve worked on. For those interested, a more extensive discussion into how machine learning can help tackle the climate crisis can be found here.

Deepening our understanding of our climate

We still don’t fully understand the climate and we need to focus on better understanding it to realise where we can best apply our limited resources. One example of climate adaptation, called climate informatics, uses big data and machine learning to classify and detect extreme weather events [8].

Climate adaptation — better predicting extreme weather events

“Spaces where domain experts meet software engineers and data scientists are the spaces where novel solutions to reducing these negative externalities can evolve.”

Where to put our energy?

The interrelated nature of the multifaceted aspects of sustainability requires you to take a step back and view the society from a bird’s eye view. What are the biggest contributors to the climate crisis, and how can technologies be applied to them? By either improving pre-existing infrastructures or enable viable alternatives are directions we can take to create positive change [9].

Let’s look at some more examples
One vital application of AI in this field is fault detection in fossil fuel pipelines. Whilst we’re still using fossil fuels, we must ensure that they are being stored and transported safely and efficiently.

In the US, it was estimated that 2.3% of total gas emissions are produced by methane supply chains, largely due to the emissions released during abnormal operating conditions [10]. It was noted that Methane emissions of this magnitude are comparable to that of natural gas combustion. Being able to monitor and detect faults in these pipelines, not only saves valuable resources, it mitigates the leakage of a gas which traps up to 100 times more heat than CO2, contributing significantly to the greenhouse effect.

Fossil fuel pipeline running across North America

Another example using big data can be found in the 2017 UN Data for Climate Action winners [11], who analysed traffic throughout Mexico City, and then evaluated how best to introduce electric vehicle charging stations, as well as the impact of various different infrastructure policies (namely : electrifying taxis, public buses or private vehicles). The research was able to successfully identify strategic locations of EV charging locations in Mexico City, as well as quantify the relative impact of CO2 reduction by the various policy choices (3.4%, 22% and 49% respectively). This kind of work is vital in enabling effective business and government decision making.

Climate change is an urgent global priority. We believe the private sector, in partnership with policy leaders, must take bold steps.” — Urs Hölzle, SVP, Google [12]

Sustainability Leadership

In the 31 years since the founding of the International Panel on Climate Change (IPCC), a global community of climate science experts giving recommendations to governments on climate policies, we have seen no significant reduction in greenhouse gas emissions. Ultimately we will need a systematic change, inspired by questioning the dogma of eternal economic growth. Sustainability demands that we consider the multifaceted aspects (environmental, societal & economic) of our industries, and ask difficult questions about how we work. Accounting in the full cost of our services and products is one aspect and also realigning our purpose towards achieving true sustainability. Codepan is now aptly positioned to develop solutions to help us transition to a more sustainable society. In the upcoming decade, we must see sustainability leadership among businesses, with measurable impact, to set the precedent for the change that is required. Authentic leadership is catalytic. Our chance to act is now.

Did you enjoy this blog post? Let us know your thoughts and requests for future articles.

Drop us a line to if you are interested in a workshop with our team

[1] UN: Sustainable development goals
[2] Data science for sustainability community SanFran
[3] Wikipedia: Jevon’s Paradox
[4] IPCC: Global Warming of 1.5 ºC (Chapter 3)
[5] NASA: Vital Signs of the Planet
[6] Umweltbundesamt: Impacts of climate change
[7] Multiple collaborators: Tackling Climate Change with Machine Learning
[8] Nature Journal: How machine learning could help improve climate forecasts
[9] Project draw down
[10] Science Journal: Assessment of methane emissions from the U.S. oil and gas supply chain
[11] UN: Data for climate action
[12] Green peace: Clickclean

Written by: William Baker Morrison

We are digital architects and technology pioneers driving the sustainability revolution [Berlin, DE]