Day 5 gave me a model that catches 81% of fraud but at a cost: 45 false alarms. Day 6 was about choosing a threshold, and it taught me that this isn't a math decision. It's a business one. The problem
Markdown supported · ⌘↵ to post
No responses yet.