ABSTRACT
In spite of the fact that a time series is an outcome of a stochastic process and, thus, an observation of a dynamical phenomena, methods, which are usually related to the analysis and modeling of static phenomena, are often applied. A class of such methods is closely related to the standard regression analysis.
The classical regression model is used to describe a static relation between a dependent variable Yt and p independent variables X1t, . . . , Xpt. In time series analysis, the observations occur successively in time and most frequently with an equidistant time distance. Therefore, an index t is introduced to denote the variable at time origin t-for other applications t denotes an arbitrary index.
