Exploratory Analysis of Gamblers’ Financial Transactions to Mine Behavioral Pattern Data

Traditional approaches to evaluate gambling behaviors have largely depended on surveys or self-reported data, which are prone to bias and subjectivity. This restriction hinders the creation of strong models to comprehend and reduce harmful gambling behaviors. This study investigates gamblers’ payment transactions to gather significant data to analyze their behavior. Studying these patterns provides important information about their spending habits, financial preferences, and overall financial well-being, which in turn leads to the development of an innovative model to reduce harms and promote “Responsible Gambling” behaviors. To meet this objective, this study employs a dataset of financial payment records collected from a digital payment provider, which serves as an intermediary between the customers’ banks and gambling merchants to identify prominent gamblers and discover the significant behavioral patterns they exhibit. Our analysis reveals the dataset follows a Pareto distribution with a high concentration of transactions within a small group of gamblers. We propose and evaluate a novel approach using dynamic “session-series” instead of daily time-series to capture gambler behavior. Preliminary quantitative evidence demonstrates the superiority of session-series in identifying specific behavioral patterns compared to daily data. This research contributes to the development of data-driven methods that understand and potentially mitigate problematic gambling behaviors.