Author: Woon Kim
Type: Master’s thesis
Abstract: This study presents a set of models for predicting incident duration and identifying variables associated with the incident duration in the state of Maryland. The incident database for years 2003 to 2005 from the Maryland State Highway (MDSHA) database is used for model development, and year 2006 for the model validation. This study, based on the preliminary analysis with the Classification Tree method, has employed the Rule-Based Tree Model to develop the primary prediction model. To enhance the prediction accuracy for some incidents with complex nature or limited samples, the study has also proposed and calibrated several supplemental components based on the Multinomial Logit and Regression methods. Although the prediction accuracy could still be improved if a data set with better quality is available, the developed set of models offers an effective tool for responsible agencies to estimate the approximate duration of a detected incident, which is crucial in projecting the potential impacts on the highway network.