Empirical Study Of Chinese Airline Network Structure Based On Complex Network Theory

With the development of China’s economy and economic globalization, air transportation has beendeveloping fast. It is a new strategic task for Chinese civil aviation to develop from a large one into apowerful one. As an important resource of civil aviation transportation, airline network determines theaccessibility of air transport, network reliability and operational efficiency and shows importanteconomic and social value. Due to the expansion of the scale of civil aviation transportation, aviationnetwork will become more and more complex, and the influence on air transport would get deeper anddeeper. Thus, the construction and management of airline network is one of the essential strategicmissions and steps to build a powerful China civil aviation. The cognition of airline network structureand the analysis of affecting factors are the foundation of airline network construction management.With the help of the Complex Network Theory, we use the2010Chinese domestic airline networksamples to carry out an empirical study of Chinese domestic airline network structure from aspects asfollows: the topology structure, hub hierarchy level and anti-destruction ability. Meanwhile, the gravitymodel and Grey Clustering Theory are also applied to the analysis of the economic and social factorsand driving forces that influence the connection of Chinese airline network services. These studies arebased on large numbers of real data samples and therefore the empirical results are reliable. As a result,this research is of certain theoretical and practical significance for people to understand Chinesenetwork structure performance and the factors that affect the network structure change, to fully developairline network functions, and to direct the management of aviation network operation.First of all, this essay analyzes the basic statistical characters and the connectivity of Chinese airlinenetwork. The results show that Chinese airline network is a small-world network with degreedistribution obeying a double power-law. There exist nodes with low degree but high betweenness. Inlongitudinal comparison, while the scale of Chinese airline network accelerates, the average number ofpath length and cluster coefficient gets lower, showing that the network structure is under optimization.In Chinese aviation network, the weight and degree of the node, formed by throughput and flightdistance, have power law relationship, but the effect of degree on throughput is more than that ondistance. The node’s weight of throughput and flight distance also presents a positive correlation withthe node’s betweenness, but the correlation coefficient is smaller, showing that the position of an airportin the airline network is not definitely decided by its throughput. Through the analysis of clustercoefficient correlation, we’ve found that when the degree value is high, the cluster coefficient shows an obvious power-law distribution on degree value, which indicates that Chinese airline network appearedto form the group structure whose center is the airports with large degree, and is characterized with a"Hub-and-Spoke "network structure.Second, using the complex network-centric theory of degree centrality, clustering coefficient andeigenvector centrality as analysis tools, this paper comprehensively compares and analyzes thecentralization level of Chinese airline network based on the hub function. It turns out that China’s airnetwork shows different levels of the centric: airports in Beijing, Shanghai, and Guangzhou are amongthe top centers and those in Shenzhen, Chengdu, Kunming have become the subordinate centers.Meanwhile, we’ve found that the three top-centre airports fail to make their position of top-levelprominent. Among the top three hub airports, Beijing airport turns out to be the weakest in the functionof centrality and Guangzhou airport the strongest. In terms of regional hub function, Urumchi airportshows the most powerful function and airports in Kunming, Chengdu, Xi’an follow. The centralizedlevel of regional hub airports in Shenzhen and Chongqing is outstanding, and so are the locations. IfGuangzhou airport could consolidate its function of national top-level hub, the vision that Shenzhenairport becomes the area hub in South China would be just in sight. Also, Chongqing airport andChengdu airport might be the area hub in Southwest China. There is still huge capability for the areahub airports to develop, which allows China’s airline network to achieve more developments.Thirdly, with the help of the principles of gravity model and the grey clustering method, wedetermine the main social factors of economy which influence the connection of China’s airline network.These factors are the urban per capita disposable income out of the airport, the town population and thecity’s tertiary industry output. A further analysis reveals that the population in the airport city does nothave obvious correlation with the node degree value of airport nor passenger throughput. However,comparatively, urban population is more correlated to the node degree value of airports and passengerthroughput than city permanent population is. It seems to result from factors like the various ranges ofChinese administrative divisions, airport radiation limits, urban residents’ purchasing power for airtransport demand and so on. The average DPI of urban residents in airport cities and the tertiary industryoutput value have strong positive correlation with the passenger throughput and the course volume: thehigher the former ones are, the geater the latter ones turn out. When we use the tertiary industry outputvalue as the primary factor in connecting the modeling of aviation network, its degree distribution turnsto be similar as that in a real network, which means the main driving factor that influences Chineseaviation network connection lies in the development level of the tertiary industry output value, while, asto their demand for air transport, the DPI just shows people’s purchasing power instead of an effectivedemand. Lastly, according to the degree value and the betweenness, we can confirm the importance of theairport nodes, and after the selective attack, we can figure out the amplitude of variation of threeindexes: the network effectiveness, the size of the biggest group and the clustering coefficient, whichwould be used in estimating the airport’s anti-worst-situation ability. Choosing9important airportnodes of Chinese aviation network as the objectives of attack, we firstly make disposable attack toevery single node, and the evaluation data show that regional hub airports contribute more to theconnectivity of Chinese aviation network, which means they are the key points to maintain thesmoothness of the network; secondly, we make continuous attack, and the figures show that only if wetake down the important airports as many as5%of the total number, can we decrease the efficiency ofthe whole network by more than50%, and when the attacked nodes increase to more than22%, theefficiency would decrease to0, which in all means being faced with deliberate attack, Chineseaviation network becomes extremely fragile. Attacks sorted by betweenness do more damage to thenetwork than attacks sorted by degree value, meaning that compared with high degree value nodes,high betweenness ones bring more influence to the connectivity of the network. With the breakdownof3hub airports: Beijing, Guangzhou and Shanghai, the passenger flow of Hohhot, Harbin andGuiyang goes up; With the breakdown of regional hub airports, the passenger flow of Lhasa, Xiningand Lanzhou goes up; when Shenzhen airport is attacked, the flow mainly goes through Beijing,Guangzhou and Shanghai to keep the network connectivity, although, it has no big effect on all theother regional hub airports.