Case Study: U.S.-Based Payment Systems Company achieves real-time fraud detection and $1M/month savings per 1% reduction with H2O.ai

H2O for Real Time Fraud Detection

U.S.Based Payment Systems Company, which processes billions of dollars in payments each month, faced the challenge of detecting rare, costly merchant fraud in real time. To improve prediction on large, complex transaction data, the company engaged H2O.ai and its H2O Deep Learning capability for scalable, real-time fraud detection.

H2O.ai implemented a distributed in-memory H2O Deep Learning solution, allowing the customer’s data scientists to run rapid test-and-learn experiments over 160 million records with 1,500 features and deploy the best-performing models for real-time scoring. The improved detection translated into clear business value: the company estimates that a 1% reduction in fraud delivers about $1 million in savings per month.