The structural evolution of the Chinese aviation network during and after the pandemic: A machine learning-based approach

Abstract

To identify the evolution pattern of the Chinese aviation network before, during, and after the pandemic, we develop a machine learning-based framework to analyze the network's development dynamics. By integrating link prediction algorithms into this framework, we quantify the contributions of 11 topological features driving structural changes. Utilizing aviation passenger flow data from China from 2014 to 2024, we identify important topological features that reveal the impact of the COVID-19 pandemic on the air network evolution. The empirical findings yield the following insights: (1) Targeted investments in core hub airports should be prioritized, given their critical role in maintaining network connectivity and facilitating rapid recovery during disruptions. (2) Airlines should strategically optimize shared connectivity and resource allocation to maintain critical routes and network resilience during times of resource constraints caused by the pandemic. (3) To control possible cascading effects caused by disruptions on international routes, secondary hubs and regional routes can be promoted. This would stabilize domestic connectivity and enhance the resilience of the aviation network. (4) Post-pandemic, the diversity-driven topological features become more prominent, suggesting airlines' plan of enhancing network robustness. Policymakers should promote the development of secondary hubs and new routes, thereby improving the aviation network's resilience and reducing excessive concentration to core hubs. These findings provide practical insights for balancing centralization, regional development, and network diversification, contributing to a resilient and adaptive aviation network capable of withstanding future disruptions.