Network Econometrics is a new set of methodologies to exploit network information when undertaking econometric analysis. It has uses in both short and long-term traffic prediction. This research will test and extend its applicability for other domains, such as predicting network structure. The basic idea is to systematically develop a weight matrix for use in statistical analysis to weight the effects of neighbours on an outcome variable. Unlike primitive spatial weight matrices which look at some form of adjacency, the network weight matrix considers network structure, so complementary and competitive observations (e.g. links) are treated differently (flows on competitive links affect flow on the link in question negatively, complementary links are positive). The measure of betweenness, computed in a network reliability context, is used to determine whether links are complementary or competitive.
Transport networks possess distinct characteristics that change over time and space. For example, capacity changes spatially, but may or may not change temporally. Although it does not change over a short period of time, it may change over a long run. The betweenness depends upon the measurement method. It may or may not change temporally but must change spatially. Total system demand (e.g. regional population) changes temporally, but is uniform spatially. The proposed network econometrics approach is able to incorporate both spatial and temporal exogenous variables and should be able to be extended to forecast the evolution of transport system, using historic data about the structure of the network and the attributes of that network.
The opportunity ID for this research opportunity is 2232