Predicting Environmental, Social, and Governance Scores with Machine Learning: A Systematic Literature Review

The paramount factors in assessing sustainability and the social responsibility of business are environmental, social, and governance (ESG). Over the last few years, leveraging machine learning (ML) methods in predicting ESG scores has preserved significant recognition due to its possibility of enhancing prediction accuracy, investment practices, and decision-making processes by stakeholders, policymakers, and investors. Our research uses ML methods to synthesize recent findings from scholarly articles on ESG score prediction. The systematic literature review (SLR) elucidates diverse areas of previous literature, such as datasets, methodology, metrics, and findings. The finding of SLR illustrates the significant application of ML to predict ESG scores, especially by Random Forest. The research paves the way for practical implications for stakeholders seeking long-term investment practices. However, it also underscores the pressing need for further studies to explore the capabilities of other ML models and enhance their capabilities through integrating hybrid models or feature engineering, thereby ensuring the continued improvement and relevance of ESG score prediction.