Evaluating ESG Performance in the Energy Sector: A Data-Driven Approach Using Cluster Analysis

This paper explores Environmental, Social, and Governance (ESG) performance within the energy sector using machine learning—specifically, K-Means clustering. Given ESG’s increasing prominence in investor decision-making and regulatory mandates, this study addresses the challenge of heterogeneous ESG data by implementing unsupervised learning to classify firms based on sustainability metrics. A 10-year dataset comprising anonymized ESG scores is preprocessed, normalized, and analyzed to derive three distinct performance clusters. The analysis reveals that while governance scores are consistently high, social and environmental metrics display wide variability. Insights are provided for investors, corporate strategists, and policymakers seeking to benchmark ESG performance more effectively in the energy domain.