Sarioglu, MertSariyer, GorkemRamtiyal, Bharti2026-04-302026-04-302025-10-14978303195962297830319596392211-09842211-099210.1007/978-3-031-95963-9_462-s2.0-105020239958https://hdl.handle.net/123456789/15680https://doi.org/10.1007/978-3-031-95963-9_46This paper integrates machine learning (ML) to propose a cluster-based ESG performance prediction model. Using data from 1235 companies across 8 sectors, we first clustered companies using their environmental, social, and governance performances and predicted their grouping using financial parameters such weighted average cost of capital (WACC), systematic risk, and market risk premium. The k-means ++ algorithm was used to cluster companies, revealing notable differences in their ESG pillar scores. To predict ESG performance group, we employed Decision Tree, Random Forest, and XGBoost comparatively. Among these the best-performing algorithm, Random Forest, had 97.57% accuracy. In the next stage, feature engineering was applied to identify key financial parameters that influence ESG prediction, where weighted average cost of capital equity and market risk premium emerging as the most significant factors. These findings demonstrate the significance of financial parameters in predicting ESG performance and present implications for investors, analysts and companies to ensure alignment with sustainability and ESG goals. The proposed model demonstrates the effective integration of ML and financial data for ESG analysis and prediction.eninfo:eu-repo/semantics/closedAccessESG PerformanceClusteringMachine LearningSustainabilityFinancial PerformanceConference Object