Forecasting Parameters

Forecasting parameters control several aspects of a demand forecast, such as the handling of missing values, outlier detection, fit and forecast validation, and sparse data forecasting.

You can improve the default settings for forecasting parameters based on your data analysis and forecast results.

The settings for all forecasting parameters related to time are in days and automatically adjusted for aggregate forecasting time levels. For example, the default setting of 182 days for a forecasting parameter is automatically adjusted to 26 days for a weekly forecast.

Note: You can’t make any change to the displayed forecasting parameters or their values in a predefined forecasting profile. You can make such changes in only a copy of the predefined forecasting profile.

This table lists the commonly used forecasting parameters.

Forecasting Parameter	Description
FillMissingMethod	Specifies how to fill any undefined historical values. Parameter values may be 0, 1, 2. 0 for no missing values field. 1 for linear interpolation based on not missing neighbors. 2 for omitting missing values.
GlobalAllocationPeriods	Specifies the number of days going backward from the history end date that's used for calculating the average demand (for bookings, consumption, or shipments).
DiscontinueDateMeasure	Specifies the measure that contains the product discontinuation date for use in the functionality for managing product launches. If the measure is specified, the forecast ends on the product discontinuation date instead of the plan's forecast end date. If the measure date is after the forecast end date of the plan, the forecast end date is used. When the selected measure doesn’t contain a value or isn’t present in the plan’s measure catalog, the forecast end date of the plan is used. If the date for the LaunchDateMeasure forecasting parameter is later than the date for the DiscontinueDateMeasure forecasting parameter, both dates are ignored, and the forecast is generated from the plan's forecast start date to the plan's forecast end date. The default value for the DiscontinueDateMeasure forecasting parameter is the Final Discontinuation Date measure. The available measures for selection are of the Date data type, are available in the forecasting profile’s work areas, and have all dimensions of the forecasting profile's output measure except for Time.
EnableNaiveForecast	Specifies whether naive modeling is used, and if so, what type. The parameter value may be 0 or a positive integer. Use 0 to disable naive modeling. 1 to use Oracle proprietary naive modeling. Any integer greater than 1 to use the simple moving average with the value controlling the number of historical periods used.
IntermitCriterion	Specifies the lowest percentage of zero values in historical demand for which the time series are evaluated using intermittent forecasting methods.
LaunchDateMeasure	Specifies the measure that contains the product launch date for use in the functionality for managing product launches. If the measure is specified, the forecast starts on the product launch date instead of the plan's forecast start date. If the measure date is earlier than the forecast start date of the plan, the forecast start date is used. When the selected measure doesn’t contain a value or isn’t present in the plan’s measure catalog, the forecast start date of the plan is used. If the date for the LaunchDateMeasure forecasting parameter is later than the date for the DiscontinueDateMeasure forecasting parameter, both dates are ignored, and the forecast is generated from the plan's forecast start date to the plan's forecast end date. The default value for the LaunchDateMeasure forecasting parameter is the Final Launch Date measure. The available measures for selection are of the Date data type, are available in the forecasting profile’s work areas, and have all dimensions of the forecasting profile's output measure except for Time.
WriteFit	Specifies whether the forecast is generated for the future or also for the history. Enter 0 to write the forecast for only the future, or 1 to write the forecast for the future and history. When you enter 1, the fit forecast is written for the entire history used during the forecasting process.
DetectOutlier	Specifies whether an attempt should be made to detect and smooth outliers in the time series. If extreme highs and lows are acceptable in your demand pattern, you can disable this forecasting parameter by entering zero.
OutlierSensitivity	Specifies the sensitivity of outlier detection. The greater the value, the more liberal the detection. For common detection, use values less than 2.
RemoveExtremeOutlier	Specifies whether aggressive outlier smoothing should be performed, and if extreme values should be removed before being processed by forecasting methods. Enable this forecasting parameter by entering 1 only if there's a clear cause to remove extreme values, if extreme values are rare or because of data errors, and you want a very smooth forecast.
EnableFitValidation	Specifies whether to enable statistical fit validation. 1 to enable validation. 0 to disable validation.
EnableForecastValidation	Specifies whether to enable statistical forecast validation. 1 to enable validation. 0 to disable validation.
FitValidationSensitivity	Controls the sensitivity of fit validation. Forecasting methods with a mean absolute percentage error (MAPE) greater than the specified value are rejected. The smaller the value, the stricter is the validation. For loose validation, use values between 1 and 2. For strict validation, use values between .3 and .5. This forecasting parameter may be the most important one for your forecast. If your forecast is failing at the required level, increasing the value of this forecasting parameter might help.
ForecastValidationSensitivity	Specifies the sensitivity of forecast validation. The smaller the value, the stricter the test. For loose forecast validation, use values between 5 and 10.

Additional forecasting parameters are available, and you can include them in your user-defined forecasting profile by clicking Actions > Add to open the Add Parameter to Viewable List dialog box. Select the forecasting parameter you want, and click the Add button to include it in the forecasting profile.

For information about forecasting parameters, refer to the white paper titled "Demand Management Forecasting Parameters" that's available in Document ID 2551481.1 on My Oracle Support.