Machine Learning in Financial Performance Prediction: The Role of ESG

Abstract

This study explores the application of machine learning techniques to predict corporate Environmental, Social, and Governance (ESG) scores, with a particular focus on identifying the most influential factors derived from company reports. Three predictive models - linear regression, random forests, and gradient boosting - were employed to estimate ESG risk scores. The experimental results demonstrate that the gradient boosting model outperforms the other approaches in predictive accuracy. Analysis using Shapley Additive Explanations (SHAP) reveals that industry classification is the most significant determinant of ESG scores, followed by key financial indicators such as Price/Sales ratio, Price/Book ratio, and Market Capitalization. The proposed predictive framework offers valuable insights for investors and corporations, facilitating informed investment decisions and strategic enhancements in ESG performance.