Development of Commuting Route Choice Model by Using Neural Network

Abstract

In this study, the route choice behavior is analyzed by using the result of the route choice survey. The neural network model, which is explained the route choice behavior, is developed. The state of congestion of the route, the number of signals and stops, the easiness of driving, and the difference of the travel time are used as input data. The model is trained with 140 samples of the survey. The percent of fitness between the estimated result and the observed one is 93%, and this result is satisfied.