Mining Spatial-Temporal Travel Patterns from Highway Transaction Data

Highways are playing increasingly significant roles in connecting different cities, and there are many vehicles traveling on highways. However, with the ever-growing number of vehicles, traffic problems such as congestion have become more and more severe, and the efficiency of travel is dramatically affected by the on-road jams. Therefore, it is necessary to understand the basic modes of highways and to identify vehicles' travel patterns, which can help the highway administrators to master the characteristics of highways and make effective decisions to relieve traffic problems. There has been much research about urban travel patterns while little research has been conducted on highway travel patterns. As far as we know, this is the first time to analyze highway travel patterns using highway transaction data. In this paper, we propose to use a practical data-mining method. Firstly, we introduce a 4-dimensional vector to describe the spatial and temporal characteristics of a car. Then we adopt the unsupervised clustering-based methods to mine the hidden regularities of the travel patterns, and we get four identifiable patterns that could be explained reasonably in practice. Finally, we provide some case studies to demonstrate the effectiveness of the proposed scheme and conduct a series of analyses based on different patterns.