A hybrid suitability mapping model integrating GIS, machine learning, and multi-criteria decision analytics for optimizing service quality of electric vehicle charging stations

Abstract

Electric vehicles are emerging as sustainable transportation solutions worldwide. Inadequate electric vehicle charging stations (EVCS) hinder their broader adoption. Optimal EVCS site selection is vital, requiring multicriteria decision-making (MCDM) analyses and geographic information systems (GIS). The research introduces, for the first time in site selection problems, an innovative methodology that integrates GIS, machine learning, and MCDM, effectively mapping the suitability of EVCS in urban environments. This study aims to fill the gap in evaluating EVCS placement in densely urbanized areas by adopting a retrospective approach to examine both primary and secondary criteria at existing EVCS sites. Focusing on Prague - a city with a dense EVCS network - it assesses their suitability using various MCDM techniques, representing a significant advance in optimizing EVCS distribution. Spatial analysis facilitated criteria reclassification, and the random forest (RF) algorithm identified key criteria, particularly transportation infrastructure and population density. Analytic hierarchy process (AHP), fuzzy AHP, and stepwise weight assessment ratio analysis (SWARA) are employed to derive criteria weights and suitability maps. Comparative results showed a predilection towards fuzzy AHP over other MCDM methods for modeling suitability analysis for placing EVCS, indicating its marginal effectiveness with the largest high-suitability area (172 km 2 ) and hosting the most EVCS (461) in this zone with the highest average score (4.49). This study not only assesses criteria importance and technique efficacy but also signifies a paradigm shift in MCDM from subjective to objective, data -driven decision-making by incorporating machine learning. The introduced approach offers guidance for EVCS planning and expansion by pinpointing areas that optimize service quality.