The Final Forecast MAPE / Accuracy and Bias page displays the forecast accuracy and BIAS data for each product-customer combination. You can change the key performance indicator (KPI) on display by using the context selector in the top-right of the page.

This selector gives options from a parameters list of five different options: C-Var Mean, C-Var Trend, C-Var Residual, MAPE/Accuracy (%), or Bias (%). 

Worksheet

The worksheet displays the C-Var and forecast accuracy measure selected in the context selector for the product-customer combination.

You can use this function before history correction. If you select C-Var (Trend) in the context selector dropdown (the C-Var measure generally used within History Correction):

  • The worksheet displays all the C-Var values.
  • You can use a filter on the worksheet to flag instances with large variances which require more focus to improve forecast accuracy. 
  • You can filter for variances above a specific threshold.

Effective Parameters

Parameters are set by a System Administrator at the global level and apply to all product-customer combinations. The parameters are shown in the right-hand panel as Effective Parameters.

Parameters

ParametersValues
Forecast Error Basis

The calculation for forecast error is the absolute forecast error divided by the selected divisor. This parameter specifies the divisor used in the calculation.

  • Actuals - divides the forecast error by history.  
  • Forecast - divides the forecast error by the forecast.  
  • Minimum Error - divides the forecast error by the either history or the forecast (whichever will give the lowest error).  Using this measure can help to prevent incentivising behavior that can introduce bias into forecasts.   
MAPE / AccuracyDefines whether to display the Mean Absolute Percentage Error (MAPE)or Forecast Accuracy (defined as 100% minus MAPE). 
Lead Time OffsetSpecifies the number of planning periods to be used as lead time offset for forecast error calculations.  For example, for a forecast made for Week 1, selecting a Lead Time Offset of 2 measures the forecast made in Week 1 from Week 3. 
Periods to Sum

Specifies the number of planning periods to sum in the calculation of MAPE.  For example, for a forecast calculated in Week 1, selecting a Lead Time Offset of 2 and Periods to Sum of 3 measures the forecast made in Week 1 offset to Week 3, and then summed for Week 3, Week 4 and Week 5.

This parameter references the Initialization Periods set in 2000 Stat Parameters: D2200: Stat Optimization Parameters The test window references the initialization period set for the forecast method and discounts the number of periods when producing a forecast. If the number of tests is set to 10, and the initialization period is set to 6, then the forecast will be based upon the difference between these two values, 4. 

Number of TestsSpecifies the number of forecast calculations to be tested.