I Trained Six Models for Fraud Detection, and the Best One Isn't in Production
Towards Data Science
Read full post
A final-year project developed NairaShield, an AI fraud detection system trained on two distinct public datasets merged into a unified schema. Despite training six models, the deployed model isn't the best performer, reflecting project constraints and regulatory workflow additions.
