Study · 02 — Lab
Anomaly detection in production orders
A hybrid pipeline that flags quantity and price entry errors before they reach the factory floor.
The problem
A typo in an order is discovered once the item has already been made.
What I built
A per-product Z-score rule, Isolation Forest and content-based duplicate detection, evaluated on precision and recall.
What changed
The pipeline flags suspicious orders before manufacturing starts.
Tools
- Python
- Scikit-Learn
- Isolation Forest