Rule based prediction of fastest paths on urban networks

Estimation of fastest paths on large networks forms a crucial part of dynamic route guidance systems. The present paper proposes a statistical approach for predicting fastest paths on urban networks. The traffic data used for conducting the statistical analysis are the input flows, the arc states or the number of cars in the arcs and the different paths of the network. The statistical method proposed is called hybrid clustering and consists of four methods namely multiple correspondence analysis, k-means clustering, Ward's hierarchical agglomerative clustering and canonical correlation analysis. The results obtained from hybrid clustering on the traffic data are decision rules that yield the fastest path for a given set of arc states and input flows. These decision rules are stored in a huge database for performing predictive route guidance. Whenever a driver arrives at the entry point of the network, the current arc states and input flows of the network are searched in the database to extract the corresponding decision rule for finding the fastest path. When no rule is found in the database for a given set of input flow and arc states, the shortest path is predicted as the fastest path.