On the application of probabilistic route choice models to urban rail transit networks containing small-scale OD trip data

Abstract

Modeling passenger route choices is crucial for analyzing and predicting public transportation demand. One of the most popular methods is to use probabilistic route choice (PRC) models (also known as discrete choice models in general), which have broad applications in transportation, economics, politics, and other fields. However, its performance varies depending on the characteristics of the origin–destination (OD) trip data and should be examined carefully. This paper proposes a framework for validating the PRC model on its application to urban rail transit (URT) networks containing small-scale OD trip data. The concept of small-scale data is defined at first for each OD pair considering the desired confidence level and the variance of route choices. Then, a travel time range (TTR)-based method is put forward to deduce passengers’ actual route choices as a benchmark for verifying PRC models. The difference and regularity analysis between the actual route choices and the model predictions are also performed with a twofold comparison. A case study on the Nanchang metro in China shows that the actual daily passenger volumes on routes of small-scale OD pairs diverge remarkably from the estimations of the PRC model. The PRC model’s performance is further discussed when the small-scale OD trip data accumulate to a larger scale over multiple days (e.g., several months). This study reveals the inherent limitation of PRC models in estimating the travel behaviors of passengers in a small-scale population. Several practical implications are discussed to improve the route choice model and passenger flow analysis.