Predictive Modelling for Esg Risk Levels in S&P 500 Companies: Empowering Informed Investment Decisions

Abstract

ESG is significant because in the world of Finance and investments, as several socially conscious investors, use ESG criteria to screen and take informed decisions related to long-term investments in a corporation. ESG metrics are not usually part of mandatory financial reporting, though companies are increasingly making disclosures in their reports. This paper studies the fundamental role of ESG factors in corporate performance evaluation, in particular companies within the S&P 500 index. The research paper is focused on developing a predictive model and its capability to evaluate and predict the ESG risk levels given a wide dataset sourced from Kaggle using advanced machine learning methods. The methodology includes exhaustive data preparation, exploratory data analysis, feature engineering, and the application of several classification algorithms such as Random Forest, Decision Tree, and Naïve Bayes. Model evaluation metrics using accuracy, precision, recall, F1-score, and AUC-ROC were applied to ensure the robustness and reliability of the prediction. The main finding of this analysis is the potential data leakage when the "ESG Risk Percentile" feature was removed and it improved the balance in the model. This contribution will underline the importance of including ESG consideration within investment practices and derive useful insights for investors and stakeholders. These findings underline the practical value of ESG risk analysis to drive more sustainable and ethical business practices, hence leading to more informed and resilient investment decisions.