Unmasking Risky Habits: Identifying and Predicting Problem Gamblers Through Machine Learning Techniques

Area of Science:

• Computer Science
• Psychology
• Behavioral Science

Background:

• Machine learning (ML) is established for identifying problem gamblers.
• Current ML methods often depend on self-reported data like account closure or self-exclusion.
• A gap exists in unsupervised, real-time identification of problem gambling behavior.

Purpose of the Study:

• To develop a novel ML approach for unsupervised identification of problem gamblers.
• To create real-time prediction models for at-risk players.
• To provide insights for interventions promoting responsible gambling.

Main Methods:

• Unsupervised learning techniques to generate labels for problem gamblers.
• Development of predictive models for real-time user identification.
• Validation of the combined unsupervised and supervised ML approach.

Main Results:

• Successfully generated unsupervised labels for problem gamblers.
• Developed accurate real-time prediction models.
• Demonstrated the efficacy of the novel combined approach.

Conclusions:

• The proposed method offers a viable alternative to self-reported labeling for problem gambler identification.
• Real-time detection enables timely and targeted interventions.
• This approach has significant potential for promoting responsible gambling and healthier player habits.