Detection Method for Communities in Railway Demand and Flow Network

The railway demand and flow network is constructed in this study, and the detection method for communities in its structure analysis is put forward by the integrated combination of several methods. First, a spectral bisection analysis is used with a normal matrix to get vector forms and basic forms of possible communities. Next, the fuzzy c-means clustering method is used with the improved Mahalanobis distance under weighted multi-properties as a similarity index to create different community combinations. Finally, the extended modularity function is used to judge and choose the best community combination. The result of the example of a network with 309 main passenger stations and 2925 train services shows that this detection method for communities is effective in structure analysis.