Research Projects

Predicting Dividend Statuses of Thai Listed Companies Using Machine Learning

This study addresses the dividend puzzle by utilizing machine learning techniques to forecast dividend status of 515 non-financial companies listed on the Stock Exchange of Thailand (SET). By using random forest, boosted tree, and decision tree models, we classified dividend status into 4 categories: grow, maintain, decline, and remain unpaid. The results indicate that the boosted tree method provided superior predictive power compared to the random forest and decision tree methods. Predicting dividend policy in emerging markets is valuable for investors and analysts to understand future dividend prospects, which is a critical signal of a company's financial well-being.