How good are network centrality measures? Longitudinal analysis of traffic in a railway network in the United States
Transportation networks are plagued by frequent delays, which pose crucial challenges to the economic advancement of a country. In previous studies, transportation networks have been envisioned as network structures of vertices with links representing a connection between two vertices. In the present work, the focus is on the delay analysis using tools of complex network analysis of the Amtrak railway network in the US. The Amtrak delay data for a period of 6 years between 2009 and 2014 was analyzed. It was observed that delay distribution varied every year and also for same stations in different years. The effect of conventional network measures on the average daily departure delay at various stations and also on the total daily traffic load on the station were estimated. Contrary to the predictive power of network topology measures in model transportation networks, it was observed that the topology measures had a negligible effect on Amtrak delay, with stations that faced the highest traffic experiencing shorter average delays. The results of this study call for additional realistic network measures and routing schemes, which could capture the features of the real-world transportation networks.
PACS Nos 89.40.-a