This paper develops a KNN Model weighted by the importance of characteristic variables of the Random Forest, to predict the travel time between two adjacent signalized intersections. The importance of each characteristic variable is calculated by the Gini coefficient evaluation index based on the Random Forest model, and is weighted into the KNN model to predict the travel time. In the case study, the density and impact of traffic lights are selected as characteristic variables due to their close relationship with travel time. The travel times are clustered by the DBSCAN algorithm to distinguish the number of stops affected by the traffic lights. Experimental results demonstrate that the proposed model provides an effective approach for urban travel time prediction and outperforms the considered competing methods. Combined with the Dijkstra's Algorithm, the proposed model is applied to the road network to find the shortest travel time path.

CICTP