Balancing Efficiency and Spatial Equity in Sustainable Electric Vehicle Charging Infrastructure: A GIS-MCDA and Machine Learning Suitability Framework for Türkiye

Transport decarbonization through electric mobility depends not only on how many charging stations are deployed but where, and whether expansion balances accessibility, grid readiness, land-use protection and regional equity. Türkiye, targeting net-zero by 2053 with electric car sales exceeding 10% of the market in 2024, shows a highly uneven charging network: provincial provision ranges from 9.0 to 155.6 points per 100,000 inhabitants, with the least-served half of the population holding only 22.3% of installed capacity (Gini = 0.311). This study develops a GIS-based multi-criteria framework treating charging expansion as a sustainability-constrained planning problem. Six criteria, namely population, GDP, transformer and transmission-line proximity, road-network proximity, and city-centre proximity, were harmonized to a 100-m grid via fuzzy membership functions, with an exclusion mask protecting sensitive land uses. Three weighting scenarios were compared: equal weights (EVCSI-A), Random Forest-derived weights (EVCSI-B), and expert AHP weights (EVCSI-C). Road accessibility (41.12%) and economic capacity (29.84%) dominated existing placement, explaining ~71% of feature importance, stable across algorithms and bootstrap replicates. National results reveal an efficiency–equity trade-off: EVCSI-B concentrates suitability in metropolitan corridors, EVCSI-A preserves broader coverage, and EVCSI-C reinforces metropolitan bias. Central and Eastern Anatolia remain underserved. We recommend sustainability-constrained screening followed by grid-capacity verification, positioning EVCSI as a transferable equity-monitoring tool supporting SDG 7, 9, 11 and 13.