Many large data sets have a network dimensionality (social networks, peer-to peer file transfer, bitcoin, organisation, supply chain, connected systems etc.). Practical applications of network analysis include understanding its resilience to disturbances, robustness to attacks, and efficiency of processes on networks (information, energy transfer). The techniques draw knowledge from applied statistics, applied physics, graph theory, and data science. More recent advances have developed graph neural networks to discover graph features to quantify challenging behaviour.
This study will be to apply feature-based graph analysis to flight data from July 2008 to demonstrate multi-scale graph analysis of UK, USA, Australia and China. The aim is to represent important notions in a low data dimension to enable effective communication and dissemination.
The objectives of this study will be to:
- Transform 1 chosen month of flight data into network form and visualise it.
- Implement graph analysis across 3 different scales:
- macro-scale (Statistical analysis),
- meso-scale (community analysis),
- node-level (centrality analysis).
- Assess and discuss the 3 different graph analysis scales and what they mean for air transport networks in real socioeconomic and engineering terms.