Willingness to pay for electric vehicle charging: A comparative study of machine learning and ordered logit models

This study analyzes factors influencing users’ willingness to pay (WTP) for electric vehicle (EV) charging in Tehran, Iran, where cheap fossil fuels and limited infrastructure pose key challenges. Data were collected from a stated preference survey of 366 residents in summer 2024. An ordered logit model (OLM) was applied to identify influential variables, and machine learning models (Random Forest, XGBoost, Support Vector Machine) were tested for predictive accuracy. Results show that WTP is highest at commercial centers and workplaces or educational institutions, about 60% higher than cost parity with fossil fuels. Longer charging times reduce WTP, with each additional minute lowering the chance of choosing higher payment options by 0.4%. Environmental awareness and the presence of amenities increase WTP. Machine learning models outperform OLM, with Random Forest reaching 75.73% accuracy compared to 59.02%. Expanding charging facilities in workplaces and educational institutions could maximize WTP, improve accessibility, and support investment recovery.