IExponentialSmoothingParameters

Assembly: Stat;

Namespace: Prognoz.Platform.Interop.Stat;

Description

The IExponentialSmoothingParameters interface determines the exponential smoothing parameters.

Inheritance Hierarchy

IExponentialSmoothingParameters

Comments

Selected trend type and seasonal effect determine whether one or another parameter is to be used.

Smoothing result depends on the parameter alpha. The more is alpha, the stronger is influence of actually observed values (when alpha=1, theoretically smoothed values of the previous period are completely ignored), the less is alpha, the stronger is influence of theoretically smoothed values (when alpha=0, actual values are completely ignored).

The gamma parameter is called growth coefficient. If gamma is included into exponential smoothing model, the trend additive of modeling series is recalculated at each step regardless of its previous value and is adapted to smoothed series changes at this step. When Gamma parameter equals to zero, the additive is not influenced by changes of smoothed series (and the source series as well), in case of linear or exponential growth it is constant for each time moment; in case of linear damped growth it decreases under influence of phi. When gamma = 1, the growth coefficient depends only on the current changes of the smoothed series.

Delta parameter is responsible for effect of source series deviation from the series smoothed during the previous period. The closer delta is to 1, the stronger is the effect.

Phi parameter determines to what extent the current growth coefficient is weakened with respect to the previous one. The more is Phi, the stronger is the coefficient.

Properties

	Property name	Brief description
	Alpha	The Alpha property determines value of alpha parameter.
	Delta	The Delta property determines value of the delta parameter.
	Gamma	The Gamma property determines value of the gamma parameter.
	Phi	The Phi property determines value of phi parameter.