Abstract:

The prediction of airport traffic demand is an important basis for the decision-making of airport construction scale. Based on a large amount of sample data, this article uses model construction and empirical analysis methods to study the prediction of airport traffic demand. Firstly, two prediction methods, correlation regression and principal component regression, are proposed to construct four types of segmented prediction models for the passenger and cargo throughput of each airport: linear regression, correlation regression, principal component regression, and stepwise regression. Empirical analysis is conducted on the constructed model using passenger and cargo throughput data from 203 airports. The empirical analysis results indicate that there are varying degrees of correlation between airport passenger and cargo throughput and various macro variables, among which there is a high correlation with variables such as gross domestic product (GDP), total retail sales of consumer goods, and residential disposable income. From the perspective of the predictive performance of the four types of prediction methods, the constructed correlation regression and principal component regression models perform well in predicting performance. Statistics result have found that under the allowable maximum absolute error percentage, over 50% of the 203 airports can be effectively predicted by regression analysis method. In addition, the higher the level of airport passenger and cargo throughput, the more effective the regression analysis method can predict the number of airports. The proportion of predictable airports with passenger throughput of over ten million and cargo and mail throughput of over 50 000 tons can both reach over 90%.

Key words: airport, passenger throughput, cargo and mail throughput, prediction method, regression analysis method