Machine Learning for Fraud Prevention

Join Ryan for an inside look at Vodafone’s Fraudulent Order Evaluation process – an enhanced fraud detection framework. Discover how we are implementing machine learning modelling with ClearScape into our proactive detection processes, while consolidating 30+ detection modules and 80+ attributes into a single streamlined workflow. Learn how we combine ML models with Risk-Based Logic to blend domain expertise with data-driven predictions, enabling more accurate and explainable fraud decisions. We’ll also discuss techniques used across the modules, such as decay scoring to prioritize recent behavior, TF-IDF and Vector Distancing for fraudulent webchat detection, and pattern-based anomaly detection to flag unusual spikes. Finally, see how our simplified architecture makes it easier to add, update and deploy fraud models at scale.