Development of a Hybrid Sustainability Index for Smart Cities Using Machine Learning and Multi-Criteria Decision Making

Abstract

This paper presents HYSI-SC (Hybrid Sustainability Index for Smart Cities), a novel framework that integrates machine learning-derived indicator weights with multi-criteria decision making (MCDM) aggregation methods to produce a dynamic, empirically grounded composite sustainability score. HYSI-SC employs a stacked ensemble of gradient boosting regressors and random forest models to learn the predictive importance of 47 sustainability indicators across environmental, social, economic, governance and technological dimensions from longitudinal panel data and feeds these data-driven weights into three MCDM aggregation methods — TOPSIS (Technique for Order Preference by Similarity to Ideal Solution), VIKOR (VlseKriterijumska Optimizacija I Kompromisno Resenje) and CRITIC-ELECTRE — to produce index scores that are robust to methodological choice. The framework is validated on longitudinal panel data from 24 Indian smart cities (from the Ministry of Housing and Urban Affairs Smart Cities Mission) and 18 UK cities (from the CDRC Urban Observatory network) over a 10-year window from 2013 to 2023. Empirical evaluation demonstrates that HYSI-SC achieves a mean absolute error of 2.3 index points against expert-assessed sustainability benchmarks, outperforms six established composite indices including the IESE Cities in Motion Index and the Arcadis Sustainable Cities Index and produces rankings that exhibit 78.4% agreement with UN-SDG-aligned expert panel assessments whilst providing objective, reproducible scores free of expert subjectivity bias.