Abstract
This article explores the adoption of alternative fuel vehicles (AFVs), leading to decarbonization, in disadvantaged communities (DACs) by applying statistical and explainable artificial intelligence (XAI) techniques to understand the factors associated with AFV adoption in these communities. The study harnesses a unique and comprehensive database of surveys and public databases for the Puget Sound region in the United States. The XAI techniques, specifically the Extreme Gradient Boosting algorithm with Shapely Additive Explanations, provide interpretable and understandable explanations of factors associated with AFV adoption in DACs. The study findings provide an understanding of the social and economic factors and challenges of DACs. The results suggest several key factors, especially a lack of access to charging infrastructure, consumer attitudes, and income, play a substantial role in adopting AFVs. As expected, AFV adoption in DACs (12.96%) is lower than non-DACs (15.30%). More public charging stations strongly correlate with AFV adoption in DACs. Tech-oriented households in DACs are more likely to adopt AFVs compared with non-DACs. The findings also point to the significant effects of home charging facilities while adopting AFVs in DACs. The XAI results emphasize the importance of socio-economic factors in AFV adoption programs and provide insights into decision-making in DACs. This research contributes to the literature on AFV adoption and suggests opportunities for improvements in DACs transitioning to AFVs. The study findings can be used to assess the planning-level impacts of refueling or charging infrastructure in DACs while enabling DACs to benefit from infrastructure investments.
Authors’ contributions
The authors confirm their contribution to the article as follows: study conception and design: ALP, AJK; data collection: ALP; analysis and interpretation of results: ALP; draft manuscript preparation: ALP, AJK; Supervision: AJK. All the authors reviewed the results and approved the final version of the manuscript.
Data accessibility statement
The analysis utilizes data collected from three publicly available resources: the 2021 Puget Sound Household Travel Survey (PHTS) (PHTS, Citation2022), the U.S. Department of Energy’s alternative fuels data center (AFDC) (AFDC, Citation2022a), and U.S. Department of Transportation’s (USDOT) Environmental Justice databases (USDOT, Citation2023). The authors will make the codes used to analyze the data available upon request.
Disclosure Statement
No potential conflict of interest was reported by the author(s).