New Study Explores Sentiment Around Electric Vehicles, Leading to Faster Government Response and More Infrastructure


Policy makers have long been wanting to improve infrastructure needed for the adoption of electric vehicles (EV), but with mass amounts of unstructured data they have been unable to determine how charging stations are performing and where more need to be added, according to a recent study from Georgia Institute of Technology researchers.

Researchers in the Machine Learning Center at Georgia Tech (ML@GT) and the School of Public Policy combined machine learning and social science frameworks to analyze reviews from EV users to determine the overall sentiment of their charging experience. Prior to this method, this process was done by hand and took on average 32 weeks. Now it can be completed via a computer in minutes, with the potential to update in near real-time, based on the study’s findings.

The data set collected from 2011-2015 from EV user mobile apps contains more than 127,000 reviews from 12,720 charging stations and ten major EV charging networks across the United States. Due to challenges in accessing behavioral data, this level of aggregation has not previously been possible. The deep learning method achieved 84.7% accuracy when compared to human analysis of sentiment.

Even though EV’s are more cost effective and have a positive impact on public health by helping to reduce air pollution, researchers found that many consumers still have a difficult time transitioning to driving an EV in public infrastructure.

“A major challenge in people adopting EV’s is the perception that infrastructure is not present to support their vehicle, when in actuality there are rapidly expanding public charging points for these vehicles at popular points of interest such as workplaces, shopping centers and transit destinations,” said Omar Isaac Asensio, lead author of the paper and an assistant professor at ML@GT and Georgia Tech’s School of Public Policy.

While charging stations are more common in urban areas, the study shows that users are 12 to 15 percent more likely to have a negative experience at an urban located station instead of a rural station. The study also found that users have similar experiences at private and government run charging stations.

Understanding these experiences may allow managers and employees in the technology and policy sectors to improve their products and create policies that better serve consumers, leading to greater infrastructure and increased adoption of this technology, according to researchers. By computerizing efforts to understand the pitfalls and spillover benefits of policies, government action and intervention may occur more rapidly.

Many of the existing incentives for owning an EV, such as tax rebates are being phased out. Other current policy initiatives, like installing charging at workplaces or getting a utility discount for installing charging at residences are designed to encourage adoption. But this doesn’t account for people in the U.S. living in multi-family residential buildings which often prevents them from having access to at-home charging, according to the study.

“In the city of Atlanta, all new residential building construction is required to pre-wire the building for EV charging. That’s a new policy that has been enacted on the local ordinance level that can really help accelerate adoption,” said Asensio.

The study, Real-time Data from Mobile Platforms to Evaluate Sustainable Transportation Infrastructure, was published in Nature Sustainability and featured on the June 2020 cover.

For more information on ML@GT, visit

Press Contact:

Allie McFadden

Communications Officer

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.