Sequential Logit Approach to Modeling Customer-searching Decisions of Taxi Drivers

Abstract

This paper develops an enhanced sequential logit model to depict the customer-search behavior of vacant-taxi drivers. This model considers that vacant-taxi drivers can change the choices they make on their way to a designated district. Trip data from the global positioning systems (GPSs) of 460 urban taxis were extracted to calibrate the model and verify the factors underlying the drivers’ search decisions. The findings reveal that the proposed sequential logit model is capable of predicting the search paths of vacant-taxi drivers. This model form is considered more informative for policymakers who aim to study search paths and the associated traffic congestion contributed by taxis in each district.