In recent years, the idea of using a mathematical model to describe the behavior of physical phenomena has been very much considered. Specifically, a definitive model, based on physical laws, enables researchers to calculate the number of time dependencies precisely at any moment in time. However, in the real world, we often face time-dependent phenomena with many unknown or unavailable factors (Lindley, 2010; Roulston et al., 2003). In this case, when it is not possible to achieve a definite - model, the prediction methods are wide used, especially when the past observations of a variable and primary relationships between specific observations are available. Forecasting methods that are used in different fields of science can be categorized based on various aspects. For example, the prediction methods used in the field of wind energy can be divided into four categories of 1) ultra short term (several seconds to four hours), 2) short term (4 to 24 hours), 3) medium-term (1 to 7 days), and 4) long term (more than 7 days) (Zack, 2003; Soman et al., 2010). Also, the structure of forecasting methods can be divided into two types of 1) single methods and 2) hybrid methods. Each of these categories can also be subdivided into smaller subgroups.