Naive Bayesian Classifier for Incident Duration Prediction

When choosing the appropriate response to an incident, it is important to predict the potential impact of the incident, including its duration, as accurately as possible. Therefore, we develop a probabilistic model based on a naïve Bayesian classifier to assist with prediction of incident duration. Two significant advantages of this model are its ability to readily accommodate incomplete information, or information received at different points in time, both of which are characteristic of the incident management process. This model was calibrated and tested using incident records from the Georgia Department of Transportation, and is shown to perform favorably compared to a standard linear regression model, as well as under a variety of information provision scenarios.

• Corporate Authors:

  Transportation Research Board

  500 Fifth Street, NW
  Washington, DC  United States  20001
• Authors:
  • Boyles, Stephen
  • Fajardo, David
  • Waller, S Travis
• Conference:
  • Transportation Research Board 86th Annual Meeting
  • Location: Washington DC, United States
  • Date: 2007-1-21 to 2007-1-25
• Date: 2007

Language

• English

Media Info

• Media Type: CD-ROM
• Features: Figures; References; Tables;
• Pagination: 11p
• Monograph Title: TRB 86th Annual Meeting Compendium of Papers CD-ROM

Subject/Index Terms

• TRT Terms: Incident detection; Incident management; Information dissemination; Information technology; Linear regression analysis; Operations; Traffic congestion; Traffic flow; Traffic incidents
• Uncontrolled Terms: Bayesian analysis; Incident duration; Probabilistic models
• Geographic Terms: Georgia
• Subject Areas: Administration and Management; Data and Information Technology; Highways; Operations and Traffic Management; Planning and Forecasting; I72: Traffic and Transport Planning;

Filing Info

• Accession Number: 01049548
• Record Type: Publication
• Report/Paper Numbers: 07-1801
• Files: TRIS, TRB
• Created Date: May 21 2007 1:18PM