Perceptions of Electric Vehicle Adoption Through Natural Language Processing and Machine Learning

The US transportation sector, responsible for 27% of GHG emissions, primarily from light-duty vehicles, is in need of more sustainable practices that decrease such emissions. Electric Vehicles (EVs) emerged as promising alternative offering advantages such as heightened energy efficiency and diminished pollution. Although the benefits of electric vehicles seem to be automatic, there are still barriers that limit their full adoption. One of the key challenges hindering EV adoption revolves around the limited understanding of the obstacles consumers perceive. Traditional survey methods, often employed to collect data on potential limiting factors, prove to be inefficient due to their high costs, time-consuming nature, and protracted data collection periods. To overcome these limitations, we propose an innovative approach that involves the integration of Twitter data, the US Census Bureau's American Community Survey, and information on US charging station locations. This study focuses on three NSF ASPIRE cities: Indianapolis, Salt Lake City, and El Paso. Insights were extracted using Natural Language Processing and Machine Learning techniques, including VADER, LDA, Feature Selection, Regression, and Classification models. The findings for this research indicate correlations between EVA and factors including non-English languages spoken at home, age group (35-44), longer commute times, and alternative transportation means. This work could support the decision-making process in how to broadcast the EVA message to communities with diverse backgrounds and develop strategies for the adoption of EVs.