The global textile and apparel (T&A) industry faces critical pressure over its environmental impact, producing approximately 92 million tons of waste annually and contributing to nearly 20% of global water pollution. For Bangladesh, one of the leading apparel exporters, the integration of transparent Environmental, Social, and Governance (ESG) practices is crucial for meeting international sustainability standards and maintaining global competitiveness. This study proposes a machine learning-based method to analyze ESG practices among LEED-certified T&A factories in Bangladesh. Leveraging contextual Natural Language Processing (NLP) and Topic Modeling technique, we developed a robust Machine learning (ML) framework based on Non-Negative Matrix Factorization (NMF) for topic extraction and a Random Forest classifier for ESG category prediction. We achieve an accuracy of 87% and an F1-score of 0.87 in ESG category prediction on a validated expert-defined keyword set, surpassing the traditional ESG analysis approach. Our analysis identified four key dominant ESG themes: Environmental Sustainability, Social I: Workplace Safety and Compliance, Social II: Education and Community Programs, and Governance. The results also show that 46% of the factories prioritize environmental initiatives such as energy conservation and waste management, 44% focus on social aspects, including safety and education at work, while governance practices are still significantly underrepresented, only in 10%. Overall, our framework offers a scalable, data-driven approach for analyzing corporate sustainability disclosures, providing actionable insights for industry stakeholders, policymakers, and global brands pursuing responsible sourcing in Bangladesh’s T&A sector.
Magotra et al. (2026) studied this question.