Page 372 - 360.revista de Alta Velocidad - Nº 6

Yang, Haoran. Dobruszkes, Frédéric. Wang, Jiao’e. Dijst, Martin.

3.2.2 Hierarchical cluster analysis (HCA)

It should be noticed that each transportation network could be composed of multiple clusters
and multiple subgroups as community networks. Community networks refer to city nodes that
are gathered into several groups in which there is a higher density of city-pair connections
within groups than among groups. HCA is a community detection algorithm based on modularity
proposed by Newman and Girvan, (2004). The basic concept of the HCA algorithm is to evaluate
the result of the network partitioning, which computes the difference between the number of
links within communities and the expected number.

We define Q as the modularity value; the higher the value of Q, the better the community
structure is. n denotes the total number of communities in the network, L is the total number of
passengers in the transportation networks, l is the total number of passengers in the community
m
m,d is the total number of cities in community m.
m
4. Results

4.1 The comparison of city strength between HSR and airline networks

4.1.1 City strength

Figure 3 The city strength of HSR (left) and airline (right) networks.

As shown in Figure 3, 29 of 105 HSR cities and 37 of 168 airline cities are dominant (that is,
DIT ≥1), and 76 and 130 non-dominant, respectively (DIT <1). Regarding the dominant cities,
Beijing, Shanghai and Guangzhou in the east are the top three cities in China but belong to

370 360.revista de alta velocidad