Supply Chain Demand Forecasting
Impact40% fewer stockouts
Stockouts-40%
Inventory Cost-15%
Stores200+
Overview
Developed a demand forecasting system for a retail chain to optimize inventory levels across their store network. The system predicts demand at SKU-store level accounting for seasonality, promotions, and local events.
The Challenge
The retailer was experiencing frequent stockouts on popular items while holding excess inventory on slow-movers, leading to lost sales and write-offs.
The Solution
Built hierarchical forecasting models using Prophet and LSTM networks, with automatic promotion lift detection and weather-adjusted predictions. Integrated with their ERP for automated replenishment signals.
Key Results
40% reduction in stockouts
15% decrease in carrying costs
Deployed across 200+ stores
Tech Stack
PythonProphetTensorFlowAirflowSnowflakeTableauSAP Integration