The forecast strategy determines how forecast values will be calculated.
All forecast strategies are based on statistical forecast procedures and, therefore, on forecast models that mathematically qualify the period of historic data. The exponential smoothing methods (exponentially weighted moving average) are currently the most widely used time series methods (see Exponential Smoothing).
If you expect historic values to continue to develop as they have in the past, choose a forecast model that fits the previous trend well.
The automatic model selection strategy allows you to let the system select the forecast model that best fits the trend of historic data (see Automatic Model Selection).
The following forecast strategies are available:
The forecast value is calculated from the medial average of the historic values.
Optional forecast parameters: Outlier Correction, Setting Negative Forecast Values to Zero, Logging Statistical Key Figures , Ignoring Initial Zeros
The forecast value is calculated according to the order.
· Obligatory forecast parameter: Order of Moving Average
The order of the moving average is a number N that determines the length of the time interval for calculating the average. This is the number of chronologically sequential historic values. The forecast value is calculated as the average of the last N historic values.
Do not enter a negative number for the order.
· Optional forecast parameters: Outlier Correction, Setting Negative Forecast Values to Zero, Logging Statistical Key Figures , Ignoring Initial Zeros
To calculate the moving average, each historic value obtains a defined weight in the weighting group.
· Obligatory forecast parameter: Weighting Group Moving Average
The weighting group for the moving average is a key that specifies
¡ how many historic values are included in the calculation of the forecast value
¡ what weight the individual historic values carry in the calculation
By clicking on Weighting Groups you will come to a screen for displaying and maintaining weighting groups.
· Optional forecast parameters: Outlier Correction, Setting Negative Forecast Values to Zero, Logging Statistical Key Figures , Ignoring Initial Zeros
Simple exponential smoothing is suitable if the historic data displays a horizontal trend.
· Smoothing factor settings: Alpha (Basic Value)
· Optional forecast parameters: Outlier Correction, Setting Negative Forecast Values to Zero, Logging Statistical Key Figures , Ignoring Initial Zeros
The forecast is calculated according to Holt’s method and is suitable if historic values display a rising or declining trend.
· Smoothing factor settings: Alpha (Basic Value), Beta (Trend Value).
· Optional forecast parameters: Outlier Correction, Trend Dampening, Setting Negative Forecast Values to Zero, Logging Statistical Key Figures, Ignoring Initial Zeros
Choose this strategy if your historic values display seasonal fluctuations (for example, annual fluctuations) from a constant base value.
· Obligatory forecast parameter: Periods per season
· Smoothing factor settings: Alpha (Basic Value), Gamma (seasonal components).
· Optional forecast parameters: Outlier Correction, Setting Negative Forecast Values to Zero, Logging Statistical Key Figures , Ignoring Initial Zeros
The forecast is calculated according to Winter/Holt’s multiplicative method and is suitable if historic values display seasonal fluctuations from a rising or declining trend. Here the extent of the fluctuation depends on the strength of the trend.
Ice cream sales in summer: It is taken for granted that ice cream sales rise by a trend of 10% annually. A seasonal increase of 30% each summer then leads to ever greater absolute fluctuations.
· Obligatory forecast parameter: Periods per season
· Smoothing factor settings: Alpha (Basic Value), Beta (Trend Value), Gamma (seasonal components).
· Optional forecast parameters: Outlier Correction, Trend Dampening, Setting Negative Forecast Values to Zero, Logging Statistical Key Figures, Ignoring Initial Zeros
The forecast is calculated according to Winter/Holt’s additive method and is suitable if historic values display seasonal fluctuations from a rising or declining trend. Here the extent of the fluctuation is independent of the strength of the trend.
An example of this is a product, 1000 units of which are sold every July over a long-term trend.
· Obligatory forecast parameter: Periods per season
· Smoothing factor settings: Alpha (Basic Value), Beta (Trend Value), Gamma (seasonal components).
· Optional forecast parameters: Outlier Correction, Trend Dampening, Setting Negative Forecast Values to Zero, Logging Statistical Key Figures, Ignoring Initial Zeros
Simple linear regression (ordinary least squares).
· Optional forecast parameters: Outlier Correction, Trend Dampening, Setting Negative Forecast Values to Zero, Logging Statistical Key Figures, Ignoring Initial Zeros