All of the models developed in this book so far have dealt with situations that arise when observations are independent of each other. The example of overseas visitors explains a time series in which the observations are dependent on the previously observed data. In a brief discussion of this example, it was established that it is necessary to develop time series models. Since the time series is sequential in nature, the time stamp may be displayed in nanoseconds, seconds, minutes, hours, days, or months.

This chapter will open with a quick review of the important concepts of time series in autocorrelation and partial autocorrelation functions, as well as fitted model assessment measures. Much like the classification and regression models, a host of methods are available for analyzing time series data. An important class of time series models in seasonal decomposition includes LOESS (STL), exponential smoothing state space models (ets), Box-Jenkins...