Enterprise Case Study|Waffarha

E-Commerce Fraud Detection

PythonAWSMySQL

The Challenge

Waffarha Has a detected Fraud detection team for supervision and customer protection. ran on people, not systems. A dedicated team manually inspect transactions to protect customers and merchants Roughly 9 hours of review per cycle. As order volume grew, the queue grew with it, fraud detected late or even missed causing issues with affiliated banks and unsatisfied customers while the process couldn't scale without hiring more reviewers.

The Outcome

Reaching 97.17% Recall, 97.17% Precision, 97.17% F1 score, 99.94% accuracy (tuned to break-even point) Production for the first month's accuracy 99.9%, 3 missed frauds out of 40K correct Transaction, and 180 frauds detected

Multiverse Support

Have a quick question? Drop us a line below.

AI

Multiverse Agent

Online & Ready

Hi! I am the Multiverse AI Agent. Ask me anything about our software engineering, AI capabilities, or the Multiverse Academy!