Objective

Review how completed reforecasts performed against demand that had already arrived.

Environment

  • Your reporting access includes the Forecasting tab.
  • Reforecast runs completed during the period you want to review.

Resolution

  1. Open Reporting and select Forecasting under Planning.
  2. Set the completion-date range and any queue filter. The queue filter selects runs that covered a selected queue; each displayed score still covers the run’s full queue scope.
  3. Read Average reforecast accuracy as the mean of stored scores from completed runs. Each run counts once, regardless of its volume.
  4. Review Reforecast accuracy per completed run and the run table to locate changes. Check Queue scope, intervals evaluated, actual volume, and change versus the previous scored run.
    The report combines stored reforecast scores with the initial forecast backtest for context. Forecast names, queue scope, volumes, and accuracy values shown here are synthetic. Select the image to inspect it at full size.
  5. Use Initial forecast accuracy only as context. It averages selected-model backtest scores for distinct parent forecasts and uses a different evaluation window.
    Note: Different evaluation windows

    Initial forecast accuracy is a historical model backtest; reforecast accuracy measures elapsed intervals at run completion.

  6. Open a run for more detail if an outlier needs investigation. Its current comparison can differ slightly from the stored score because more actuals may have arrived since the run completed.

N/A means a run had no stored accuracy score and is excluded from the average and chart. A score of 0% means the evaluated miss exceeded the actual volume; it is a measured result.