Comparative Study of RP/SP Combined Data Estimation Between Mixed Logit and Nested Logit Model

RP/SP combined model is very important for the travel behavior research. Mixed logit and nested logit model for RP/SP combined data estimation are compared and three main results are put forward in this paper. Firstly, sometimes the correlation between similar transportation modes is more important than RP and SP data. Different Nested logit model is necessary for classifying the data. Secondly,mixed logit model,which takes the heterogeneity of individual into consideration and treats the parameters as random variable, can deal with the correlation both between similar transportation modes and the RP and SP data. Its estimation result is better than that of the nested logit model. Finally, mixed logit model can more distinctly show the different sensitivity of different people to the travel time and monetary cost. It can reflect a fact that car users have higher value of time than public transportation users.