Research on the classification characteristics of Chinese airports from the perspective of network

Abstract

In order to accurately grasp the classification characteristics of Chinese airports, scientifically and effectively support Chinese airport division and slot capacity assessment, it is urgent to conduct a systematic and scientific analysis of the classification characteristics of airports. From the perspective of the network, the concepts of airport weight value and airport SIR value are introduced by using formal concept analysis and propagation dynamics method. Combined with the characteristics of network topology, airport support resources and airport operation resources, 17 potential categorical features on airports are proposed. Taking the main coordination, auxiliary coordination and non-coordination airports currently classified in China as labels, a machine learning classification model is established by using the random forest algorithm. Finally, the classification effect is evaluated based on the accuracy of the model and the features are ranked by importance. The research shows that the classification accuracy of the random forest model has reached 95.28%, and the classification effect is significant; the airport weight value is an important factor affecting the airport category; the classification feature proposed based on the network perspective is the key to affect the airport classification, which accounts for 46.6%, the classification features based on airport operation resources are second, accounting for 28.2%, and the classification features based on security resources account for 25.2%; the proposed method provides theoretical support and reference for the scientific division of my country's airports and the promotion of airport slot capacity assessment.