Analyzing ESG Cluster Characteristics of Thai Equity Market Sectors using Unsupervised Machine Learning
Abstract
This study presents the application of unsupervised machine learning techniques to analyze and cluster the Environmental, Social, and Governance (ESG) performance characteristics of companies listed on the Stock Exchange of Thailand (SET) that received the 2024 SET ESG Ratings. The methodology employs Principal Component Analysis (PCA) in conjunction with K-Means Clustering for dimensionality reduction and to group companies based on similarities in their ESG attributes. The study analyzed data consisting of environmental, social, and governance indicators from 186 companies. The results indicate that the companies can be categorized into three distinct primary groups: 1) A group excelling in environmental performance; 2) A group focused on transparency in ESG data disclosure; and 3) A group distinguished by social and governance performance, particularly in internal organizational management. Furthermore, the clustering results were found to be consistent with the SET ESG Ratings levels and the industry characteristics of the companies in each group. This study demonstrates that unsupervised machine learning using the K-Means algorithm can effectively explain organizational ESG operation patterns within the Thai context and can be utilized to support the development of long-term corporate sustainability strategies.
Authors: Paweenuch Tanapornwattana, Bunthit Watanapa, Chakarida Nukoolkit