Investigating the Role of Data Quality and Diversity in Improving Payment Fraud Detection Models: An Exploratory Study

Abstract

Payment fraud detection is a critical challenge for businesses and financial institutions, as fraudulent activities lead to significant financial losses and undermine trust in digital payment systems. While various fraud detection models have been developed, their effectiveness heavily relies on the quality and diversity of the data used for training and validation. This exploratory study investigates the role of data quality and diversity in improving payment fraud detection models. By examining different dimensions of data quality, such as completeness, accuracy, and timeliness, and exploring the impact of data diversity in terms of transaction types, customer demographics, and geographical coverage, this study aims to provide insights into enhancing the performance and generalizability of fraud detection models. The findings highlight the importance of data preprocessing, feature engineering, and dataset curation in building robust and effective fraud detection systems.