Explainable artificial intelligence-driven sustainability for electric vehicle charging station siting: A hybrid LIME-SHAP for MCDM
International Journal of Applied Resilience and Sustainability · 2025
DOI: 10.70593/deepsci.0101001 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Applied Resilience and Sustainability, 2025. Verfasst von Dimple Ravindra Patil und 3 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref).
Abstract (Originalsprache)
Electric Vehicle Charging Station (EVCS) planning typically relies on subjective factor weighting, hence expert bias may be introduced in multi-criteria decision problems. We propose an Explainable Artificial Intelligence (XAI) powered enhancement to a GIS–MCDM siting by relying on LIME and SHAP hybrid-based approach for inducing data-driven Multi-Influencing Factor (MIF) weights of the weighted overlay. This method applied to four wards in Mumbai India, combines global (SHAP) and local (LIME) model-agnostic importances from a balanced surrogate classifier trained using spatial samples around observed EVCS locations. The hybrid weights substitute the MIF prior inter-relation and drive the classical overlay and TOPSIS stages, leaving intact interpretability and auditability. Validation shows the degree of enhancement in site discrimination (ROC–AUC = 0.846) relative to that of the baseline MIF–TOPSIS process (ROC–AUC = 0.826) with more separated high- and low-suitability classes with less affectedness responding to single-factor perturbations, which can be attributed to the benefit of XAI-based weighting on these weights. It is expected that this will give rise to a more reliable and replicable map of EVCS suitability which can enhance overall sustainability benefits and transparent, stakeholder facing decision-taking.
Zitieren
APA
Patil, D. R., Rane, N. L., Koneti, S. B., & Rane, J. (2025). Explainable artificial intelligence-driven sustainability for electric vehicle charging station siting: A hybrid LIME-SHAP for MCDM. International Journal of Applied Resilience and Sustainability. https://doi.org/10.70593/deepsci.0101001
Vancouver
Patil DR, Rane NL, Koneti SB, Rane J. Explainable artificial intelligence-driven sustainability for electric vehicle charging station siting: A hybrid LIME-SHAP for MCDM. International Journal of Applied Resilience and Sustainability. 2025. doi:10.70593/deepsci.0101001
BibTeX
@article{Patil2025,
author = {Patil, Dimple Ravindra and Rane, Nitin Liladhar and Koneti, Subramanya Bharathvamsi and Rane, Jayesh},
title = {Explainable artificial intelligence-driven sustainability for electric vehicle charging station siting: A hybrid LIME-SHAP for MCDM},
journal = {International Journal of Applied Resilience and Sustainability},
year = {2025},
doi = {10.70593/deepsci.0101001}
}
EUVIMED