Forecast error (WAPE) and bias
WAPE is the total deviation of forecast from actual; bias shows whether the method overstates or understates.
Formula
WAPE = Σ|actual − forecast| ÷ Σactual; bias = Σ(forecast − actual) ÷ Σactual
How to read
Computed on past data, so it is an estimate, not a guarantee. For seasonal apparel the error is higher than for basics: check it on your own past seasons.
Where it is discussed
Topics
- Replenishment and stock cover
- Forecast and budget
- Stock accuracy, shrink and availability
- Range: width, depth and price tiers
- Capital, multi-brand and the planning cycle
- Season results: brand, category, SKU
- KPI dictionary and common definitions
- Data model of standard reports
Calculations
- Intermittent demand: Croston and SBA for rare sizes and accessories
- Safety stock from forecast error
- Sales forecast: method, horizon, error
- Holt-Winters: a forecast with trend and seasonality, error on held-out periods
- Forecast value added by stage and the price of error in roubles: the season assortment
- Checking an AI forecast against a naive baseline: WAPE, bias, improvement, share of periods and the gain net of costs
- Forecast of a new item from analogs
- S&OP: forecast bias, error and the balance of demand and supply
- Forecast accuracy by week: window, bias and forecast value added
Documents
- Budgeting methods: rolling forecast, ZBB, what-if scenarios
- Plan versus actual and profit factor analysis (bridge)
- S&OP: cycle and documents
- Descriptive statistics for retail: center, spread, outliers, intervals
- Demand forecasting: smoothing, seasonal indices, analogs and error control
- Holt-Winters, seasonal decomposition and Croston: forecasting with trend, season and intermittent demand
- AI in fashion and analytics: where to apply it, where not, how to check the result, which data are needed and which errors occur. A memo and a checklist
- Forecast accuracy and the value of an AI forecast: WAPE, bias and FVA on a season assortment