The potential of an econometrics approach for forecasting air transport demand is examined. The aim is to identify and select the most significant variables for multivariable linear regression models. Each independent variable has been evaluated by means of ARIMA models after rendering the historical series stationary. The variables for the different cases examined have been chosen using Student's t-test and the correlation matrix to determine their level of significance and of correlation. The resulting models have been tested on three airports in Sardinia (Cagliari, Olbia and Alghero) in order to identify the most suitable and characteristic variables for representing air transport demand. Time series for annual passenger movements have been constructed for each airport expressed in absolute terms and the index number. For the covering abstract see ITRD E128239.
Abstract