Forecast Accuracy & Forecast Value Added: MAPE, WAPE, Bias & FVA
Measure demand forecasts like a planning team: errors, MAPE vs WAPE vs sMAPE, bias and tracking signal, MAE/RMSE by SKU and family, the naive benchmark, an FVA stairstep from statistical to consensus forecast, ABC/XYZ targets and an accuracy dashboard, in 8 hands-on labs on one consumer goods company.
The first lesson is free — no account needed.
About this course
A forecast accuracy course that measures one fictional consumer goods company's demand forecasts the way a planning team does, for demand planners, S&OP analysts and supply chain students. You calculate errors, absolute and squared errors and the MAE for one SKU, compare MAPE, WAPE and sMAPE across ten SKUs and watch two tiny SKUs push MAPE to an absurd level, measure bias as a percentage with a running tracking signal and a ±4 warning rule, compare MAE and RMSE at SKU and family level, test the statistical forecast against the naive forecast and calculate its forecast value added, build an FVA stairstep from the statistical forecast through planner overrides to consensus, set realistic targets by XYZ class and finish with an accuracy dashboard. FVA is credited to Michael Gilliland of SAS, and the course is not affiliated with SAS. The first lab is free.
Lessons
- Forecast Errors: Error, Absolute Error & Squared ErrorFree
- MAPE vs WAPE vs sMAPE: Why MAPE Breaks on Small Actuals Pro
- Forecast Bias, Bias % & the Tracking Signal (with a Warning Rule) Pro
- MAE & RMSE by SKU and by Product Family: Accuracy Depends on the Level Pro
- The Naive Benchmark & Forecast Value Added (FVA) Pro
- FVA Stairstep: Statistical Forecast vs Planner Override vs Consensus Pro
- Accuracy by Segment: ABC/XYZ Classes & Realistic Targets Pro
- Forecast Accuracy Dashboard: Accuracy %, FVA, Bias & a Trend Chart Pro