Estimation Of Nested Logit Model Of Urban Residents’ Travel Behavior Using Bootstrap Method

Nested Logit model is an important disaggregate model.In order to grasp the travel rules of urban residents,predict the traffic volume reasonably,plan the urban layout and coordinate the urban traffic development,it seems important to estimate the parameter of Nested Logit model accurately.Nowadays,the main estimation methods of Nested Logit model are the simultaneous maximum likelihood estimation method and the sequential estimation method.However,the simultaneous maximum likelihood estimation method is too complicated and the sequential estimation method has the bias in the estimation of the variances that yields overestimation on the t-value.Therefore,the bootstrap method is applied to estimate Nested Logit model,because it can reduce the bias and be more effective.In this paper,the bootstrap method was introduced,including its basic idea,the common bootstrap method and its application of regression analysis;then,the activity-based travel demand theory was introduced,the individual travel behavior characteristics were ignored for four-step method and the disaggregate model can make up for this flaw,the utility maximization theory and the travel utility function of disaggregate model were introduced,at the same time,some common Logit models were introduced.Nested Logit model of urban residents’ travel behavior was established and R language was used to estimate the parameter;finally,based on the part of 2007 person trip survey data on Tokyo metropolitan area,the bootstrap method and the sequential estimation method were used to estimate the parameter.As the result is shown,it can prove that using bootstrap method can reduce the bias and improve the effectiveness of Nested Logit model.