Forecast Accuracy
How close the demand forecast was to actual demand, usually 1 minus the weighted mean absolute percentage error.
Formula
Forecast accuracy = 1 - Sum of absolute errors / Sum of actual demand
Absolute error = |Forecast - Actual|
Worked examples
- 01Four product families, forecast 1200, 800, 450, 300, actual 1100, 860, 380, 310. Absolute errors 100, 60, 70, 10, total 240. Weighted MAPE = 240 / 2650 = 9.1%, forecast accuracy = 90.9%.
- 02Single item: forecast 500, actual 420. Error 80 / 420 = 19.0%, accuracy 81.0%.
- 03Lag matters: accuracy of the forecast made 3 months before is 78 percent, one month before 89 percent. Production needs the lag that matches its lead time.
Pitfalls
- Measuring on the total only lets errors of families cancel out.
- Not freezing the forecast snapshot, so it is compared after it was adjusted.