Energy can seem like a politically divisive issue. Democrats earnestly support renewable energy adoption as a path to reducing greenhouse gas emissions, while Republicans reject solar subsidies and prefer to frack first and ask questions later, right? Wrong. In an especially divisive election year, it may come as a breath of fresh air to learn that clean energy knows no party.
We pulled the addresses of 1.5 million Democratic and Republican party donors in the top 20 solar states and analyzed their rooftops using satellite images and PowerScout’s A.I.-based image recognition model to determine which had adopted rooftop solar.
What we found is that 3.06% of Democratic donors and 2.24% of Republican donors in these states have installed rooftop solar. When you look at the numbers by state, the numbers get even more interesting- in Hawaii for example, Republicans install more solar than Democrats! Check out the infographic below to see how all the states stack up.
It seems that in well-established solar markets like California and Hawaii, party affiliation does not matter when it comes to solar adoption. In California, Republicans and Democrats install solar at equal rates. In states with nascent solar markets like Oregon and Colorado, it looks like Democrats are leading in solar installs over Republicans.
So how did we do it?
PowerScout started with the publicly available list of donors to the Democratic and Republican parties and candidates, and we focused on the top 20 solar states. Listed addresses for donors can include work addresses, PO Boxes and apartments, so we focused on the addresses that can support rooftop solar by filtering for donors that live in single family homes. This left us with 1.5 million homes across these 20 states – 1 million Democratic donors and 500,000 Republican donors.
We then pulled the satellite images for these 1.5 million homes and processed them with a machine learning model called a convolutional neural network (CNN). If you’re looking for a more in depth explanation of convolutional neural networks, there’s a pretty good explanation here, but you can think of a CNN as a computerized version of the part of the human brain that helps us see.
We trained this model with hundreds of thousands of labeled images of homes with and without solar panels until the model was able to distinguish between the two types. Now we can process thousands of images at a time to determine which homes have solar and which ones do not. Neat!
Hope you enjoy the infographic below. If you’d like to learn more about solar for your home, check out PowerScout’s Solar Calculator.
Sources: PowerScout analysis; 2012 donation data from the Stanford Database on Ideology, Money in Politics, and Elections.
Also published on Medium.