Scholarlink

<p>This paper explores latent perceptual constructs that define private vehicle commuters’ intention to adopt EVs, using a cross-sectional survey combining stated and revealed preference data across two Indian metropolitan areas. Twenty-seven perception-based variables, based on economic, technological, infrastructural, environmental, social and policy dimensions, were synthesized using Principal Component Analysis (PCA) into five, theoretically coherent constructs. These latents were later incorporated into predictive machine learning models, including Support Vector Machines (SVM), XGBoost, k-Nearest Neighbors (KNN), and an Artificial Neural Network (ANN) of EV adoption readiness based on behavioral intention theories. SHAP-based explainability identified perceived reliability, infrastructure readiness, and policy stability as the most common factors to predict adoption intention. Factorial scenario simulations indicated substantial interaction effects and that combined policy interventions had a greater effect on non-adoption probabilities than isolated improvements. The results highlight the importance of combining dimensionality reduction, nonlinear predictive modeling and explainable AI to provide insights for evidence-based approaches to EV adoption in emerging economies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Modeling Electric Vehicle Adoption Among Private Vehicle Commuters Using Perceptual Constructs and Explainable Machine Learning for Sustainable Mobility

  • 期刊论文

  • 发表日期：

    2026年6月17日

查看全文
  • Rupam Sam,
  • Subojit Debnath,
  • Sudip Kumar Roy

摘要

This paper explores latent perceptual constructs that define private vehicle commuters’ intention to adopt EVs, using a cross-sectional survey combining stated and revealed preference data across two Indian metropolitan areas. Twenty-seven perception-based variables, based on economic, technological, infrastructural, environmental, social and policy dimensions, were synthesized using Principal Component Analysis (PCA) into five, theoretically coherent constructs. These latents were later incorporated into predictive machine learning models, including Support Vector Machines (SVM), XGBoost, k-Nearest Neighbors (KNN), and an Artificial Neural Network (ANN) of EV adoption readiness based on behavioral intention theories. SHAP-based explainability identified perceived reliability, infrastructure readiness, and policy stability as the most common factors to predict adoption intention. Factorial scenario simulations indicated substantial interaction effects and that combined policy interventions had a greater effect on non-adoption probabilities than isolated improvements. The results highlight the importance of combining dimensionality reduction, nonlinear predictive modeling and explainable AI to provide insights for evidence-based approaches to EV adoption in emerging economies.