Travel Time Estimation on a Link without Real-Time Data by Correlated Links

This paper presents a novel method to estimate travel time on a road segment using information from other road segments. This method is useful especially in the case that real-time traffic on such road segment is not available. The proposed method is based on the correlation between the road segment itself and the most related road segment. We measure the relation between road segments by dynamic time warping algorithm and apply the K-Nearest-Neighbor algorithm to select the best neighbor segment to estimate the travel time on the target road segment. We found that the best attributes set that can measure the correlation between road sections consists of location of the road segments, day of the week, and current time. The link correlation results can be used as reference data to determine the travel time on the roads that are related.