Seasonal time series model comparison for nonstationary passenger flight data

Abstract

The objective of this paper is to analyze the number of passengers flying a sample of three airlines before and after 9/11 to discover whether there has been a recovery. The three airlines were modeled using simple linear regression and time series analysis. Dummy variables and trigonometric functions were used to mimic the seasonal variation and additive decomposition was used to remove the seasonal component and model the trend. The additive decomposition quadratic models were deemed the best fits. From the quadratic models is concluded that the three airlines chosen for this paper have recovered from the effects of 9/11.