A MODIFIED PAIRED COMBINATORIAL LOGIT ROUTE CHOICE MODEL WITH PROBIT-BASED EQUIVALENT IMPEDANCE

摘要

This paper presents a modified paired combinatorial Logit route choice model that takes the advantages of both Probit and Logit models to produce more reasonable flow prediction with close-form probabilities that can be easily calculated. To better account for the observation error of the model, parameters are calculated basing that the variance of impendence is related to the measured travel time. The concept of stochastic equivalent impendence is proposed to calculate the utility of each route pair and Clark's approximation for normal distribution is used to compute the equivalent impendence. The proposed equivalence concept also eliminates the need of computing similarity index as shown in most modified Logit model. Two numerical tests on the well-known examples show that the proposed model is more close to Probit model than original PCL model with reasonable errors.