ML Fraud Detection

Machine-learning model for financial fraud detection on large-scale transaction data, built on the IBM Z platform, start to finish in 24 hours. 2nd place in London: £250, IBM LinuxONE server access, a 3-month mentorship, and an invitation to speak at a GSE UK conference.

Why it matters

A real, judged result on a real dataset: 97.68% accuracy under a deadline, against a strong field.