Development of parameters of multinomial logit models

The use of multinomial logit models for predicting mode-choice decisions for urban travel has gained prominence in recent years. The prime advantage of these models lies in their ability to allocate travel demand among many modes at the same time, as opposed to the traditional binary mode-choice models developed in the early sixties. A major impediment to the increased use of logit models is the absence of a structured and welldocumented process that the analyst may follow in order to develop the model in its entirety. The purpose of this paper is to present a systematic procedure for calibrating and validating logit models through an efficient use of available data base. A case-study application is also presented to demonstrate the applicability of the method.