Using Data Science to Make Clean Energy More Equitable
Here are some lessons inspired by a recent panel the author moderated about how data scientists can help put equity into practice.
Photo by Pedro Henrique Santos on Unsplash
Data science plays an increasing role in the fight against climate change. It can be used to help with everything from predicting supply and demand for clean energy systems to identifying which buildings to target for energy efficiency upgrades. As we use data science to develop innovative solutions to climate change, we should be careful not to replicate, or worsen, existing inequities - or create new ones.
The impacts of climate change, from extreme weather to debilitating drought, will hit harder in poor countries and communities least equipped to deal with the consequences. Unfortunately, many of the solutions we are developing are also inequitable. Poor households that still cook with coal and kerosene can’t afford solar roofs and energy-efficient appliances. Low-income parents juggling work and childcare have little time to spend at charging stations to charge electric vehicles and may not have a garage or driveway to charge them at home.
As we use data science to help fight climate change, we should look for solutions that are both greener and more equitable. For example, many communities don’t have access to reliable public transportation. According to Dr. Tierra Bills, Assistant Professor of Civil and Environmental Engineering at UCLA, data on transportation needs is often unavailable or inaccurate, especially for groups like the elderly or disabled who face access issues. New geolocation data sources and methods for collecting transit preferences directly from users can offer a more comprehensive picture of people’s transportation needs. Data science could help us optimize route planning and create on-demand bus routes and carpool matching services for underserved areas.
Another area where data science can support more equitable solutions is access to clean, reliable energy. A staggering 760 million people worldwide lack electricity and many more experience frequent disruptions. According to energy expert Dr. Xin Ma, Managing Director of the Asia Platform at TotalEnergies Ventures, advances in data science can help us expand and democratize access to energy. For example, microgrids and mini grids can bring energy to remote and underserved communities by enabling small scale, localized electricity generation that is not dependent on the traditional power grid. Data science is used to forecast energy supply based on weather patterns and to monitor and optimize the condition of solar panels and batteries. In recent years, microgrids and mini grids have brought power to more than 11 million people worldwide.
These examples, drawn from a panel I moderated at last fall’s C3E conference, show how the thoughtful use of data science can support equitable clean energy solutions. On March 7, the discussion will continue in a panel moderated by my colleague Dr. Jenny Suckale at the Women in Data Science Worldwide Conference hosted at Stanford University and online. Here are some lessons inspired by these panels about how data scientists can help put equity into practice.
1) Consider the many dimensions of equity. In the past, efforts to promote equity often focused on giving people equal access to resources and opportunities. But equal access rarely leads to equitable outcomes when people start on uneven footing. Equity is a broader framework that looks in part at how benefits and burdens are distributed across different groups (race, income, gender, etc.) and geographies (rural vs. urban areas, Global North and South). It also takes people’s agency into account, asking questions like: Who is represented when decisions are made? What power, if any, do people have to change their situations and shape their communities? When looking at datasets with an equity lens, ask what is being measured, by whom, and for what purpose, as well as who is included in the data and who is not. It is also critical for marginalized groups to be represented in data science and data-driven decision making, so they play a role in decisions that will impact their communities.
2) Understand the human side of the problem you want to solve. It’s important to start any data science project with a clear articulation of your goals, rather than just gathering a bunch of data and throwing an algorithm at it. If you don’t understand the human impact of the problem you are trying to solve, you cannot develop equitable responses. For example, researchers at the Lawrence Berkeley National Laboratory were working with data on rooftop solar installations. It was only by combining this dataset with demographic data that they were able to get insight into equity questions such as who is (and isn’t) installing solar on their homes, why or why not, and what barriers stand in the way of uptake among groups with less solar penetration.
3) Look out for “data deserts” and work to correct them. If people aren’t represented in the data we have about a problem, they will be invisible when solutions are being designed. For example, if you’re developing supply and demand models for clean energy systems but don’t have data on the power needs of remote rural areas, the resulting system is unlikely to serve them well. There are useful technical approaches to address sparse or messy data, but we shouldn’t simply patch over data deserts using statistical techniques. Instead, we need to ensure data collection represents all relevant people and places. We need to think broadly about how we can get more, better data and push to open up data access across the private, research, and government sectors. This often requires carefully designed stakeholder and community engagement.
4) Acknowledge the limits of existing data. Data shows us what people do given existing resources and constraints, not what they ideally want or need or what they would do under different incentives. For example, areas with inconsistent access to power may have low energy usage not because there isn’t demand, but because people don’t install appliances or engage in activities that they might if they were guaranteed a consistent power supply. We shouldn’t assume that how people behave now is how they would prefer to act in a more equitable world.
5) Be careful not to smooth over inequities with big data. While big data is all the rage these days, it isn’t the only scale we should be looking at. Equity problems are often local and in the weeds. These small signals can get washed out when they are just a blip in a massive dataset. For example, average travel times for commuters may mask the outliers whose only option to get to work is a two-hour, multi-stage journey. You can avoid smoothing over inequities by using techniques that disaggregate data and get as granular as possible. For example, agent-based models like the BEAM model for transportation planning allow you to look at people’s behaviors and preferences at an individual level.
6) Find innovative ways to gather data on community needs. Marginalized groups are often underrepresented in existing data sources. Look for creative ways to collect both quantitative and qualitative data about these communities - and to gather their input about the data collection and analysis process itself. New data-gathering tools like mobile apps and online surveys can help, but be cautious that tech-enabled tools don’t further overlook people with limited access. It’s essential to directly engage with marginalized groups, collaborating with key stakeholders and community members to understand their perspectives on what questions to ask, what data to collect, and what solutions to design.
These are just some of the ways we can use data science to help address climate change in a more equitable way. Right now, there’s a lot of “equity-washing” happening. Organizations claim to take equity into consideration but aren’t really doing much to challenge the status quo. Instead, equity needs to be front and center throughout the data science process, from gathering data and analyzing problems to designing and implementing solutions. This is how we can build a world that is both greener and more just.
Dr. Margot Gerritsen is a professor in the Department of Energy Resources Engineering at Stanford. She is the co-founder and co-director of Women in Data Science (WiDS), an initiative to inspire and educate data scientists worldwide, regardless of gender, and to support women in the field. If you want to learn more about how to use data science to address critical issues like climate change in an equitable way, WiDS invites you to attend their annual conference on March 7, 2022 at Stanford University and online across the world.