Travel Mode Choice Modeling: A Comparison of Neural Networks and Multinomial Logit Model

In recent years,neural network has been widely applied to the travel demand prediction area,with the discrete choice model as the main analysis method,of which the prediction performance is rarely directly compared.Therefore,the inherent disadvantages of the logit model and the good characteristics of the neural network method are first analyzed in this paper.Then,with the difference and the connection of the two methods,a multinomial Logit model and a neural network model are compared for the travel mode analysis.It is found that the BP neural network model with the hidden layer can describe the nonlinear relationship between property variable,and can achieve a better prediction performance of the travel mode choice than the multinomial logit model.