Supervised vs Unsupervised Fraud Detection
This talk will cover a brief history of Co-Op Insurance success with Supervised approach to Fraud detection, and our belief that in certain use cases (fraud being one of them) that supervised models can only take you so far... Enter Unsupervised Fraud detection, including brief overview of techniques used (Isolation Forest & Auto-Encoder).
Co-Op Insurance has a rich history serving its members & customers for over 150 years. Mike Yates has 15 years experience in Analytics / Data Science across different industries including utilities, retail and insurance. He currently uses Agile practices to manage a team of Data Scientists, adding value across a variety of use cases including fraud, pricing & customer experience.