Linking Crash Occurrence to Real-time Weather Conditions and AVI Traffic Data: Ensemble Data-mining Approach

As advances in traffic detection technology help to operate roads more efficiently and as the authorities are shifting toward these advanced non-intrusive systems, the interest in incorporating safety into traffic management systems has also grown accordingly. The data that are collected from such systems is one of the greatest assets that should be utilized appropriately to maximize the benefit for the roadway authority as well as for the road users. Buried within this vast amount of data is useful information that could make a significant difference in how these roads are managed and operated. Data mining techniques are known for their superior performance in classification and prediction. This paper examines the usefulness of traffic data collected from Automatic Vehicle Identification (AVI) system, weather data and roadway geometry in real-time crash analysis utilizing data mining methods. All data mining models were found to outperform the classical ones. Artificial Neural Network and decision trees were found to provide the best accuracy in terms of the area under the ROC curve, lift chart and misclassification error for both training and validation datasets. Moreover, using data mining ensemble technique to combine the results from the best models further enhanced the prediction accuracy.

• Supplemental Notes:
  • Abstract reprinted with permission from Intelligent Transportation Society of America.
• Corporate Authors:

  ITS America

  1100 17th Street, NW, 12th Floor
  Washington, DC  United States  20036
• Authors:
  • Ahmed, Mohamed M
  • Abdel-Aty, Mohamed
  • Yu, Rongjie
• Conference:
  • 19th ITS World Congress
  • Location: Vienna , Austria
  • Date: 2012-10-22 to 2012-10-26
• Publication Date: 2012

Language

• English

Media Info

• Media Type: Digital/other
• Features: CD-ROM; Figures; References; Tables;
• Pagination: 11p
• Monograph Title: 19th ITS World Congress, Vienna, Austria, 22 to 26 October 2012

Subject/Index Terms

• TRT Terms: Automatic vehicle identification; Crash analysis; Data mining; Decision trees; Highway safety; Highway traffic control; Neural networks; Real time information
• Subject Areas: Data and Information Technology; Highways; Safety and Human Factors; I85: Safety Devices used in Transport Infrastructure;

Filing Info

• Accession Number: 01501900
• Record Type: Publication
• Files: TRIS
• Created Date: Dec 23 2013 7:53AM