Traffic Path Recommendation Model based on a Weighted Sum of Extracted Parameter

Abstract

A path recommendation for vehicular traffic is important task of traffic analysis. It is a challenging problem for researchers to extract all paths and recommend the shortest path between Origin and Destination (OD) pairs. This paper comes up with a model which is established on the weighted sum of selected link references to recommend a path for OD pairs. First, to maintain spatial dependence between link references, a vehicular traffic network of roads is proposed as a rectangular coordinate system. The algorithm based on K-means and smoothing is introduced to select link references across OD pairs. A distance aggregation algorithm is proposed to evaluate all possible paths across an OD pair. Finally, out of overwhelming paths, the algorithm recommends the shortest distance path across an OD pair. Our proposed model effectively selects the link references and gets an overall shortest path recommendation. The proposed model analyzes the non-Euclidean distance of selected link references. Our experimental analysis shows that on an average, the first four link predictions lead to 77.37% distance coverage for the recommended path.