Time Series and Stationarity
As presented earlier, time series will often have a well-defined trend and seasonal components, and a good fitting model will account for these components in such a way that the residuals will tend to have means zero, constant variance, and of course autocorrelation. On the one hand, the adjacent observations have correlations that can be positive or negative, as for example, the airline passenger data where higher values are followed by higher values. On the other hand, adjacent observations can be negatively correlated such as, for example, when higher sales values are followed by lower numbers, etc. The models to be discussed will serve as the residuals with complex autocorrelation patterns for time series regression model.