Cost-Sensitive Decision Making for Online Fraud Management

Every online transaction comes with a risk and it is the merchant’s liability to detect and stop fraudulent transactions. Merchants utilize various mechanisms to prevent and manage fraud such as automated fraud detection systems and manual transaction reviews by expert fraud analysts. Many proposed solutions mostly focus on fraud detection accuracy and ignore financial considerations. Also, highly effective manual review process is overlooked. We propose Profit Optimizing Neural Risk Manager (PONRM), a decision maker that (a) constitutes optimal collaboration between machine learning models and human expertise under industrial constraints, (b) is cost and profit sensitive. We suggest directions on how to characterize fraudulent behavior and assess the risk of a transaction. We show that our framework outperforms cost-sensitive and cost-insensitive baselines on three real-world merchant datasets.