ABSTRACT

A time series is a set of observations of the dependent variable taken specific times. Time series models seek forecasts of the dependent variable as a function of time. It is understood that there are underlying factors other than time affecting the dependent variable. There are two reasons why time series models are often used to forecast a dependent variable: the process explaining the changes in the dependent variable may not or may be extremely complicated and the ability to forecast changes in the dependent variable may be more important than understanding the process. The objective of time series methods is to determine the historical patterns in the past demand data and to forecast those patterns into the future. There are four types of patterns in the past demand data which can be distinguished by time series methods: a horizontal pattern, a seasonal pattern, a cyclical pattern and a trend pattern.