A numeric-based machine learning design for detecting organized retail fraud in digital marketplaces

Area of Science:

• Computer Science
• Criminology
• Data Science

Background:

• Organized retail crime (ORC) poses a significant threat to retailers, online marketplaces, and consumers, with increasing prevalence due to e-commerce expansion.
• Existing fraud detection research primarily focuses on financial services, leaving a gap in studies specifically addressing ORC.
• The financial and security repercussions of ORC are substantial and projected to escalate with increased internet connectivity.

Purpose of the Study:

• To develop and present a scalable machine learning strategy for detecting and isolating organized retail crime (ORC) listings on e-commerce platforms.
• To address the scarcity of research on ORC detection by contributing a novel methodology.
• To build a system capable of distinguishing fraudulent listings from legitimate ones.

Main Methods:

• A supervised learning approach was employed, utilizing historical buyer and seller behavior and transaction data.
• The framework incorporated custom data preprocessing, feature selection (45 out of 58 features), and advanced class imbalance resolution techniques.
• Classification algorithms were optimized to effectively discriminate between fraudulent and legitimate marketplace listings.

Main Results:

• The best-performing detection model achieved a recall score of 0.97 on the holdout dataset.
• The model demonstrated strong generalization capabilities with a recall score of 0.94 on out-of-sample testing data.
• The methodology successfully identified ORC listings using a refined set of 45 features.

Conclusions:

• The proposed machine learning strategy offers a scalable and effective solution for detecting organized retail crime (ORC) on online marketplaces.
• The study successfully contributes to the limited body of knowledge on ORC detection, providing a practical framework.
• The high recall scores indicate the model's potential for significantly mitigating the impact of ORC on businesses and consumers.