A Novel Approach to Extract Significant Patterns of Travel Time Intervals of Vehicles from Freeway Gantry Timestamp Sequences

Abstract

It is attractive to extract and determine the key features of traffic patterns for mitigating road congestion and predicting travel time of vehicles in traffic analysis. Based on the previous work that is a scalable approach via a Hadoop MapReduce programming model, this paper aims to extract significant patterns of travel time intervals of vehicles from freeway traffic in Taiwan, and meanwhile to compute the statistics of these patterns from the point of view one may concern. Experimental resources are the records of timestamp gantry sequences of vehicles passed in five months from 2016/11 to 2017/3 that were downloaded from the Traffic Data Collection System, one of Taiwan government open data platforms. To select one specific gantry sequence for demonstration, the longest sequence on the trip within the Taiwan National Freeway No. 5 is selected. Experimental results show that some statistics of vehicle travel time intervals according to 24 h per day are computed for illustration. These statistics can not only provide clues to experts to analyze traffic congestions, but also help drivers how to avoid rush hours. Furthermore, this work is able to handle a larger amount of real data and be promising for further traffic and transportation research in the future.

Keywords

• traffic analysis
• maximal repeat
• timestamp sequence
• Hadoop
• MapReduce