Modeling and Predicting Traffic Accidents at Signalized Intersections in the City of Norfolk, VA

The proposed research is an extension of the previously completed studies on the accident-patterns in the City of Norfolk, VA in which a multiple-regression model was developed using a selected set of intersections data from the City. The objective of this proposal is to develop comprehensive a statistical exploratory and predictive model for intersections accidents in the City. The previous models were based on a selected set of variables like driveway density, some road geometry variables such as number of lanes and safety features including median. Data on many roadway geometry factors and roadway hazards were either not collected or not available at the time of study. Also, the multiple-linear regression technique used in the previous models has presented limited predictive capability. The proposed research will improve the previously developed models by including additional variables and advanced statistical modeling. The available literature suggests road geometry and other related controllable factors influence the traffic accident rate and can be delineated using statistical methods. This research will utilize a two-step statistical analysis methodology. In the first step, exploratory statistical models will be developed from a randomly selected set of intersections in the City of Norfolk using generalized linear model (GLM) technique. These models will be validated using similar variables from the other randomly selected set of the intersections in the City. The best fit model will be proposed as the predictive model for the City. The major deliverables from the proposed research will include the following: * A validated exploratory statistical model that will include variables which provide the most valid explanation of traffic accidents. The model development process will include a set of geometrical and roadside hazard factors as independent variables. * A predictive statistical model resulting from step one could be used for accident prediction for similar road conditions in the City. The proposed research will commence in May 1, 2010 and will conclude on May 29, 2011. The major elements of the work are shown below along with the expected start and finish time for each work element. * Research Preparation: Complete literature review on the recent articles in traffic safety as well statistical modeling. (May 1, 10- June 30, 10) * Data Collection: Data collection on road geometry, road hazard and other related road variable data. (July 1, 10- Sept 15, 10) * Development of Exploratory Models: Using generalized linear modeling technique develop multiple regression models. (Sept 16, 10-Dec 31, 10) * Model Validation: Using different set of intersection validate the statistical models. (Jan 2, 11-March, 31, 11) * Select Predictive Model: Selection of best statistical model. (April 1, 11-April 30, 11) * Review of the Results and the Model: Review of the results and the statistical model. (May 1, 11-May 15, 11) * Report preparation: Prepare final report (May 17, 11- May 29, 11)


    • English


    • Status: Completed
    • Funding: $457000.00
    • Contract Numbers:


    • Sponsor Organizations:

      Research and Innovative Technology Administration

      Department of Transportation
      1200 New Jersey Avneue, SE
      Washington, DC  United States  20590
    • Performing Organizations:

      Eastern Seaboard Intermodal Transportation Applications Center

      Hampton University
      Hampton, VA  United States  23668
    • Principal Investigators:

      Dsouza, Kelwyn

      Maheshwari, Sharad

    • Start Date: 20100301
    • Expected Completion Date: 0
    • Actual Completion Date: 20110228
    • Source Data: RiP Project 26206

    Subject/Index Terms

    Filing Info

    • Accession Number: 01468195
    • Record Type: Research project
    • Source Agency: Eastern Seaboard Intermodal Transportation Applications Center
    • Contract Numbers: DTRT06-G-0029
    • Files: UTC, RIP, USDOT
    • Created Date: Jan 3 2013 3:46PM