Guy Taylor, head of data at Nedbank, reveals how a tight knit collaboration with IBM's Data Science and AI Elite team using Watson Studio built a predictive machine learning model to anticipate ATM outages and cash needs. The results? Nedbank dropped the false positive rate for fraud detection from 60–80% to below 35%. The bank also expects to see a one percent reduction in its ATM outage times per year, resulting in hundreds of thousands of dollars in savings.