Modified Betweenness Centrality for Predicting Traffic Flow

Centrality measures define the relationship between particular structural features of a network. Although network analysis has been utilized to express the most relevant physical, topological and static properties of networks, dynamic and temporal aspects are mostly disregarded. For example, in travel network studies the physical street network is assumed to be static, although it can have relevant dynamic and temporal constraints such as current traffic volume in comparison to capacity or night time closures. Traffic flow can be defined as the process of physical agents moving along an urban travel network (Kazerani and Winter 2009). Since these agents are dynamic and purposeful, they have specific travel demands such as to leave an origin or reach a destination at a specific time. If the agents consider centrality of streets in their route planning, their route planning problem becomes essentially a dynamic one. Furthermore, their time dependent demands contribute to the dynamics of centrality measures. So the research question is how to modify conventional centrality measures by considering the time dependent travel demand. So far, betweenness centrality (Freeman 1977) has been used as one of the prominent centrality measures for analyzing physical street network. In order to consider dynamic and temporal aspects of agents’ travel demand in the street network a modified version of betweenness centrality will be developed to study people’s origins and destinations in three different time periods of the day. The result will be compared to the traditional betweenness centrality of the street network to see the impact including these temporal and dynamic aspects on explaining traffic patterns.