📘
Demand Forecasting Pipeline
MediumML System Design
Design a demand forecasting system for a retail chain.
Scenario: A retailer with 500 stores needs to forecast product demand for the next 14 days to optimize inventory and reduce waste. Forecasts must be generated daily for 50,000 SKUs across all stores.
Your Task: Design the forecasting pipeline.
Key Challenges:
- Hierarchical forecasting (store × product combinations)
- Handling seasonality, promotions, and external events
- Probabilistic forecasts (not just point estimates)
- Scaling to 25M forecasts daily
Design Mode
📝 Your Design Approach
Describe your system design approach. Consider components, data flow, and key decisions.
🎯 Design Questions(Select all that apply)
Q1.Which modeling approach works best for retail demand forecasting?
Q2.How should the system handle new products with no sales history?
Q3.What evaluation metric best captures forecast quality for inventory decisions?
Q4.How should the pipeline handle promotional events and price changes?
0 of 4 questions answered