Electric Vehicles as A Sustainable Energy Technology: Observations from Travel Survey Data and Evaluation of Adoption with Machine Learning Method

Governments worldwide are promoting Electric Vehicles (EVs) to achieve the energy conservation and emissions reduction goal, but low penetration of EVs means that it still has far to go before stepping into the sustainable energy future. In addition to technological breakthroughs to enhance the appeal of EVs, how to locate demand and make targeted promotions is also vital in increasing the share of EVs. This study explored the household characteristics related to EV adoption with the household travel survey data and proposed a LightGBM-based prediction modelling framework with high accuracy and explainable results. During this process, the details of sampling techniques to overcome data imbalance were discussed with the aim of improving the model performance. Furthermore, through constructing the model with high interpretability, we identified important factors regarding EV adoptions through both statistical significance and feature importance analysis. The research findings can not only assist EV manufacturers in targeting potential buyers but also help policymakers understand families’ EV purchasing decisions and develop more targeted and equitable policies and incentive programs. Such two-pronged efforts have the potential to advance the sustainability transition towards greener transportation systems.

© 2023 Elsevier Ltd.