American Society of Civil Engineers

Data Analysis with Outlier Detection to Detect Pavement Roughness

by Shuyan Chen, (College of Transportation, Southeast University, 210096 Nanjing, China and Department of Electronic Engineering, Nanjing Normal University, 210097, Nanjing E-mail:, Wei Wang, (College of Transportation, Southeast University, 210096 Nanjing, China E-mail:, and Jian Lu, (College of Transportation, Southeast University, 210096 Nanjing, China)

pp. 2602-2607, (doi:

     Access full text
     Purchase Subscription
     Permissions for Reuse  

Document type: Conference Proceeding Paper
Part of: Logistics: The Emerging Frontiers of Transportation and Development in China
Abstract: The pavement roughness is an important index to indicate the function of the road surface. In this paper, outlier mining technique has been introduced to detect the pavement roughness by recognizing the outlier hidden in the traffic data sets collected. First, the principle, characteristic and time complexity of two typical outlier mining approaches, statistical-based and distance-based outlier algorithms, have been analyzed. Second, we applied these two approaches to pavement roughness detection, and made a comparison between statistical-based and distance-based outlier algorithms. In particular, we studied the influence of parameter k in distance-based outlier algorithm on detection results by carrying out experiments. The experimental results have illustrated that outlier mining methods are feasible and valid to detect pavement roughness, and have a good potential use in the domain of traffic engineering.

ASCE Subject Headings:
Data analysis