Demand forecasting for a retail chain

The brief

A retail chain wastes money on overstock and loses sales to empty shelves. Build a forecasting system for item-level daily demand across stores, then turn forecasts into order quantities and measure the savings. Good looks like: forecasts that beat simple baselines across many items, an ordering policy tested on past data, and a business case with the money saved.

What you will build

1. Explore sales hierarchies, seasonality, prices and events
2. Build baselines (naive, moving average, seasonal)
3. Train global machine-learning forecasts (LightGBM) with features
4. Evaluate across the hierarchy
5. Turn forecasts into order quantities and simulate stock levels
6. Write the business case and present it

Free, open data to build it with

• M5 Forecasting (Walmart sales) ↗ Kaggle
• Online Retail II ↗ UCI Machine Learning Repository

Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.