PREDICTION OF FUTURE TRAVEL TIME CHANGEFROM THE PRESENT USING STORED TRAFFICDATA ON URBAN EXPRESSWAY

摘要

A change from the present travel time to the future travel time is predicted by two steps. In thefirst step, the present travel time based on arrival is estimated more accurately by theimproved TST model, in which a unit time interval is divided into some shorter intervals. Inthe second step, the change of the future travel time from the present travel time is predictedby another model with stored traffic data, which adopts one of the Data Mining techniques,CART. Applying these models to observed data on Nagoya Expressway in Aichi, Japan, wefound that the travel time change was predicted accurately by these models even when trafficcongestion occurred.