Number of periods in a season, for example, 12PERIO = 12 if your data is monthly observed values.
Enter “X“ if you want the system to correct outlier values. For more information on outlier correction, see Outlier Correction.
If you have activated outlier correction using parameter 13OUTL, you can determine the deviation value as of which outlier correction is to be performed. Parameter 14SIGFAC is optional. 2.0 is used as the default value.
For forecast strategy 12 (moving average) it is obligatory to enter an order for the moving average.
If no value is specified or if the value is less than 2, forecast strategy 11 (simple average) is applied automatically.
For forecast strategy 13 (weighted moving average) it is obligatory to enter a weighting group.
You can determine a trend dampening factor to dampen the trend for forecast values. For more information on trend dampening, see Trend Dampening.
These parameters are used for exponential smoothing. These parameters are ignored if you have selected optimization (estimation of smoothing factors) or automatic model selection.
Alpha is used for all exponential smoothing models, beta only for those containing a trend component, and gamma only for those containing a seasonal component.
These parameters are optional. All smoothing factors have a default value of 0.3.
Enter “X“ if you want the system to estimate smoothing factors. For more information on estimating smoothing factors, see Optimizing Smoothing Factors for Exponential Smoothing.
The optimization variable defines the error measure to be used for optimizing smoothing factors. The following values are possible:
...
1. MSE (mean squared error)
2. MAE (mean absolute error)
3. MAPE (mean absolute percentage error)
4. MAD_S (smoothed absolute deviation with exponentially decreasing weights for the residuals).
Error is understood as the difference between an observed value and a forecast value.
This parameter is optional. The default value is MSE.
To restrict the search space for alpha, beta, or gamma, you can set the appropriate parameters: 23ALPHA_F and 24ALPHA_T, 26BETA_F and 27BETA_T, 29GAMM_F and 30GAMM_T. The parameters that end with “F” (from) specify the lower limits. The parameters that end with “T” (to) specify the upper limits of the search space.
These parameters are optional. The default value for lower limits is 0.0. The default value for upper limits is 1.0.
You can set initial step sizes for each smoothing factor in the optimization procedure. The search algorithm uses these step sizes in the first optimization phase in which a grid search is performed.
Therefore small step sizes mean better starting values for the direct search, but also mean increasing calculation costs. This is especially critical for models with trend and seasonal components, where all smoothing factors (alpha, beta, and gamma) are required.
These parameters are optional. The default value is 0.1 in each case.
You can specify key figures for which you want to calculate forecast values. This parameter has to be set if any key figures included in your planning level are to be excluded from the forecast. You do not need to set this parameter if forecasting is desired for all key figures.
Enter KYFNM01 = 0AMOUNT to calculate forecast values for a key figure amount. You can enter additional forecast values, for example, KYFNM02 = 0QUANTITY for a key figure quantity.
You can define characteristic values for the observed values (past periods).
Your planning data is differentiated from the observed values by means of a characteristic version 0VERSION. You want to calculate forecast values for planning on the basis of your observed values. Define 0VERSION as a parameter and enter “ACTUAL” or something similar.
As parameter names are limited to eight characters, the characteristic name may have to be shortened. For this to work, the first eight characters of the characteristic name have to be unique.
If the observed values are to be taken from another key figure, specify the name of this key figure in the parameter REFRATIO.
The forecast variable is key figure 0KEYF1. The observed values are to be taken from key figure 0KEYF2. You have to set REFRATIO = “0KEYF2” to define this reference.
In order to be able to display characteristics in the log display that describe or identify the objects for which the forecast is performed, you can enter the appropriate parameters for characteristic names (CHANM<nn>).
You want to create forecast data for every product in a region that is identified by the characteristics 0REGION and 0PRODUCT. Create parameters CHANM01 and CHANM02 and assign the values of characteristics “0REGION” and “0PRODUCT” to them.
Enter “X“ if you want the system to disregard zero values in the observed values.