Almost half of the world's population is carried by airlines each year, and understanding this mode of transport is important from economic and scientific perspectives. In recent years, the increasing availability of data has led to complex network and agent interaction models which attempt to gain better understanding of the air transport network and develop forecasts. In this case study paper, we review existing research on two key approaches, namely: (1) a top-down multi-scale network science approach, and (2) a bottom-up entropy-maximization interaction network approach. Using simple socioeconomic indicators, we were able to construct a very accurate interaction model that can predict traffic volume, and the model can forward estimate the impact of population growth or fuel cost. Using network science approaches, we were able to identify community structures and relate them to economic outputs. We also saw how hubs evolved over time to become more influential. Looking into the future, using random graph theory, it seems that reduced flight cost will lead to increased hub influence. The disseminated knowledge in this case study paper will provide both academics and industry practitioners with steps forward to co-explore the interesting research landscape.• Bottom-up entropy-maximization interaction model, which considers consumer choice; • Top-down network science analysis, which seeks to uncover common statistical patterns and infer latent knowledge.The former gives a complex and detailed understanding of how spatial networks (i.e., flights) form from spatial processes (i.e., airports) and what the weight of each edge (i.e., passenger volume) is with respect to cost (impedes flow) and benefit (attracts flow) functions that relate to consumer behaviour. The latter approach gives a statistical understanding into the fundamental network properties and how they evolve over time, enabling the application of generalized network scaling laws that can be used to predict the future structure of the network. Both the bottomup and the top-down approach is of fundamental interest to network science and industry.Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.