Forecasting Methods
In Advanced Planning Systems (APS), so-called quantitative or objective forecasting methods supported. These derive forecasts based on the analysis of demand data observed in the past. To this end, the demands are arranged in a time series.
If the demand under consideration is influenced by the development of other influence factor over time causal methods such as multiple linear regression methods are applied.
On the other hand, if no causal relationship is known, then the demand is extrapolated based on its past observations (times series methods). The time series methods used in APS are related to the structure of the demand time series to be forecasted.
Usually, the following methods are supported:
- Simple moving average
- Simple exponential smoothing (for stationary, level demand)
- Double exponential smoothing (for demand with trend)
- Holt's procedure (for demand with trend)
- Winters' procedure (for demand with trend and seasonal influences)
For sporadic demand, often Crostons's method is implemented.