The business problem
FleetPulse moves industrial parts across the Eastern US. Their planners build demand forecasts in spreadsheets — slow, inconsistent across analysts, and almost always too smooth. Stockouts happen during seasonal peaks; excess inventory eats cash during troughs. Build a forecast that captures the seasonality, ships as a tool the planning team will use, and is accurate enough to actually move the inventory dial.
This case applies Decision 7 (Time Series) at enterprise scale.
Case kit
Everything you need
- Student WorksheetFrame the forecast horizon, plan your approach, defend your method.
- Enterprise Demand DatasetMulti-year demand history at SKU × region granularity.
- Python Forecasting NotebookReference implementation with decomposition, classical, and modern forecasting.
- Excel Demand PlannerThe shipping artifact — a planner-friendly spreadsheet that ingests history and outputs the forecast.
Topics you'll be applying
- Decision 1 — Data Prep & EDA — multi-year sales data has structural breaks
- Decision 2 — Regression — covariates (promos, holidays) help
- Decision 7 — Time Series — your core technique