A GIS-Based Hybrid Travel Forecasting Mode: Model Design and Preliminary Results

Author: Jinsoo You
Organization: University of Illinois

111 T.B. Hall, 611 L. Taft Dr.,
Champaign,, IL 61820
USA

Phone: 217
Fax: 244-6564
t-kim7@uiuc.edu

The main purpose of this paper is to explore a possible way of predicting link travel times for congested road networks by implementing a hybrid forecasting model with GIS technologies. In a separate study (You and Kim, 1999b), a nonparametric regression model has been developed as a core forecasting algorithm to reduce computation time and to increase forecasting accuracy. In this study, the core forecasting algorithm has been integrated with GIS technologies to implement a hybrid travel time forecasting model. A prototype hybrid forecasting system has been developed and tested by deploying GIS technologies in the following areas: (1) storing, retrieving, and displaying traffic data to assist in the forecasting procedures, (2) integrating historical databases and road network data, and (3) building and verifying routes on road networks. This study shows that adopting GIS technologies in link travel time forecasting tasks is efficient for the two contradictory goals: reducing computation time and increasing forecasting accuracy.