Predictive Classification of Online Gambling Impacts in North Sulawesi Using Naive Bayes

General background: Online gambling has become a growing societal issue in Indonesia, affecting various economic and social dimensions. Specific background: North Sulawesi is identified as one of the most vulnerable regions, with a high volume of gambling transactions and significant social consequences. Knowledge gap: Previous studies in Indonesia have focused mainly on behavioral, legal, or health aspects, with limited use of predictive modeling to map community vulnerability. Aims: This study aims to analyze the social, economic, and psychological impacts of online gambling and to develop a predictive model using the Naive Bayes algorithm to classify impact levels. Results: Based on data from 504 respondents, the model achieved an accuracy rate of 94.05% in classifying impacts into light, moderate, and severe categories. Precision and recall were highest for light and severe categories, while moderate categories showed lower performance. Novelty: This research introduces a data-driven predictive approach within a regional Indonesian context, which has been underexplored in prior studies. Implications: The findings provide a practical foundation for regional governments to design targeted interventions, ranging from digital literacy programs to psychological rehabilitation services, based on data-driven risk classification. Highlight: The model achieved 94.05% accuracy in classifying online gambling impacts. Regional data provided structured insights into social, economic, and psychological consequences. Findings support evidence-based policymaking for targeted local interventions.