Forecast value added (FVA)
Forecast value added (FVA): how much more accurate a forecast is than simply repeating last week. If the value is below zero the forecast adds no value and a complex model or manual override only harms. The measure is used when reviewing the forecast method.
Formula
FVA = naive-forecast WAPE − forecast WAPE
How to read
Suppose naive forecast WAPE is 30% and yours 24%: FVA is +6 pp. Compare on the same weeks and items: otherwise the difference is random. A negative FVA is a reason to simplify the method.
Where it is discussed
Topics
- Forecast and budget
- Stock accuracy, shrink and availability
- Season results: brand, category, SKU
- KPI dictionary and common definitions
Calculations
- 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 accuracy by week: window, bias and forecast value added
Documents
- AI in fashion analytics and content: what can be computed, limits, risks
- 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