If you have run any pipelines, they will need to be run again after changing these settings.
This range is used to identify forecasta numerical prediction of a future value for a specified time period for each unique combination of BY variable values values that are problematic. Set the boundary below and above the historical average. Values that fall outside of the boundaries that you set are identified as problematic.
SAS Visual Forecasting identifies time seriesan aggregation of transactional data into specified time intervals and sorted according to unique combinations of the default attributes (BY variables) with almost flat forecasts, where the forecasts have minimal variation or trend over the forecast horizonthe number of intervals into the future, beyond a base date, for which analyses and predictions are made.. Use these settings to define forecasts that are close to flat, but not completely flat. You can use the Almost Flat Forecasts attribute to filtera set of specified criteria that are applied to data in order to identify the subset of data for a subsequent operation, such as continued processing. these time series in the Filters pane.
This feature was introduced in release 2026.01. See Upgrading Projects to 2026.01 or Subsequent Releases in What’s New in SAS Visual Forecasting if upgrading from a prior release.
Specify the method used to identify forecasts with minimal variation or trend. This setting determines how almost flat forecasts are detected across your time series.
These conditions are controlled by the absolute and relative tolerance settings.
Specify the minimum threshold for almost-flat detection in the same units as your forecast data. This value sets a fixed tolerance at or below which forecasts are considered to have minimal variation or trend.
The absolute tolerance provides a baseline threshold that applies regardless of the forecast scale or historical patterns. For example, if your data represents sales revenue in thousands of dollars, a value of 0.005 means that variation of $5 or less would trigger almost-flat detection.
This threshold works together with the relative tolerance setting. The detection uses whichever threshold is larger: the absolute tolerance or the relative tolerance multiplied by the reference scale (historical standard deviation or forecast mean, depending on the method selected).
Specify the relative threshold as a value between 0 and 1. This is used as a percentage for almost flat detection. For example, 0.02 is a 2% relative tolerance. This value scales the tolerance based on your data's characteristics, making detection adaptive to different forecast levels or historical patterns.
The reference scale used for comparison depends on the detection method.
This method compares the forecast standard deviation to the historical standard deviation. A value of 0.05 (5%) means that forecasts are flagged when their standard deviation is less than or equal to 5% of the historical standard deviation. This threshold works together with the absolute tolerance setting. The detection uses whichever threshold is larger to ensure appropriate sensitivity across different data scales.
This method uses two comparisons:
A value of 0.05 (5%) means that forecasts are flagged when both of the following conditions are met:
This threshold works together with the absolute tolerance setting. The detection uses whichever threshold is larger for each comparison.
These mean methods compare to the forecast mean level. A value of 0.05 (5%) means that forecasts are flagged when the variation is less than 5% of their mean level.
This threshold works together with the absolute tolerance setting. The detection uses whichever threshold is larger to ensure appropriate sensitivity across different data scales.