Peer groups that actually catch anomalies
Compare a merchant to the wrong 'normal' and the anomaly disappears. Measured: 91% of cases caught with peers built on identity, 24% with peers built on behavior.
Data science, scoped up front and delivered on the agreed date.
I build demand forecasts, anomaly alerts and data pipelines that businesses use to make decisions. Every project starts with a written scope and a delivery date.
Selected work
Demand forecasting & reorder simulation for an online retailer
SKU-level demand forecasts feeding a reorder/safety-stock simulator that makes the £ cost of forecast error visible.
How I work
- Fixed scope, in writing, before work starts. What gets delivered, and what doesn't.
- A delivery date I commit to. If something threatens it, you hear about it early, not on the day.