The following parameters are always required:
This forecast strategy determines how forecast values will be calculated. The following table provides an overview of the strategies available and which parameters have to be set in relation to which strategies.
Forecast strategies and dependent parameters
ID |
Description |
Dependent parameter |
11 |
Average |
|
12 |
Moving average |
Order of moving average 15MAORD |
13 |
Weighted moving average |
Weighting group 16GEWGR |
21 |
Simple exponential smoothing (constant model) |
Smoothing factor alpha 18ALPHA or Optimization of smoothing parameters 21OPT and optimization variable 22OPTVAR with, where necessary, search space limits 23ALPHA_F, 24ALPHA_T and initial step sizes for optimization 25ALPH_S. |
22 |
Linear exponential smoothing (trend model) |
Trend dampening factor 17TDAMP; Smoothing factors alpha 18ALPHA, beta 19BETA or Optimization of smoothing parameters 21OPT and optimization variable 22OPTVAR with, where necessary, search space limits 23ALPHA_F, 24ALPHA_T, 26BETA_F, 27BETA_T, and initial step sizes for optimization 25ALPH_S, 28BETA_S. |
23 |
Seasonal exponential smoothing |
Periods per season 12PERIO; Smoothing factor alpha 18ALPHA, gamma 20GAMMA or Optimization of smoothing parameters 21OPT and optimization variable 22OPTVAR with, where necessary, search space limits 23ALPHA_F, 24ALPHA_T, 29GAMM_F, 30GAMM_T , and initial step sizes for optimization 25ALPH_S, 31GAMM_S . |
24 |
Trend-seasonal exponential smoothing |
Periods per season 12PERIO; Trend dampening factor 17TDAMP; Smoothing factor alpha 18ALPHA, beta 19BETA, gamma 20GAMMA or Optimization of smoothing parameters 21OPT and optimization variable 22OPTVAR with, where necessary, search space limits 23ALPHA_F, 24ALPHA_T, 26BETA_F, 27BETA_T, 29GAMM_F, 30GAMM_T, and initial step sizes for optimization 25ALPH_S, 28BETA_S, 31GAMM_S. |
31 |
Simple linear regression |
Trend dampening, if necessary |
41 |
Automatic model selection |
Periods per season 12PERIO, if necessary; Trend dampening factor 17TDAMP |
The parameters for outlier correction (outlier correction 13OUTL and Sigma factor 14SIGFAC) can be combined with any strategy.
Define a variable for the time characteristic (e.g. 0FISCPER) and specify which periods the past periods should be, and thereby, what the space of time for your observed values is. The observed values are the reference data.
See also: