The use of machine learning techniques to identify problem gamblers has been widely established. However, existing methods often rely on self-reported labeling, such as temporary self-exclusion or account closure. In this study, we propose a novel approach that combines two documented methods....