Each traveler chooses a departure time for each trip which is dependent on his or her travel time prediction and these decisions form the pattern of the temporal distribution of travel demand. On the other hand, the temporal distribution of travel demand has an undeniable impact on the creation of the traffic congestion pattern in the urban transport network. Therefore, for the prediction of the congestion pattern in the urban transport network, we need to predict the departure time choice of travelers. In this study, the modelling of departure time for those trips which have time constraints in the destinations are considered. The main applications of these models are as a part of a dynamic assignment, activity scheduling in activity-based models, four-step strategic modelling and assessment of demand management policies. In this study, various multinomial logit (MNL) models on different choice sets (based on numerous clustering methods) were calibrated and prediction power of these models was compared with other methods such as using fixed coefficients and survival analysis models. The results showed the efficiency of proposed models and MNL model which is based on the K-means method had a better prediction power in this approach. On the other hand, behavioral attributes of travelers got involved in the model using structural equation modelling (SEM) and concerning the conceptual framework of hybrid choice model and prospect theory, departure time choice models were made. In these models, extraversion, punctuality and conscientiousness, as well as the value function of prospect theory as latent variables, were considered in the model. The results indicated the proposed model based on prospect theory had better goodness of fit indices rather than the model based on hybrid choice.
Keywords: departure time choice model, discrete choice, clustering, structural equation modeling, prospect theory.